Tag Archives: download

MP3 Stream Rippers Are Not Illegal Sites, EFF Tells US Government

Post Syndicated from Ernesto original https://torrentfreak.com/mp3-stream-rippers-are-not-illegal-sites-eff-tells-us-government-171021/

Free music is easy to find nowadays. Just head over to YouTube and you can find millions of tracks including many of the most recent releases.

While some artists happily share their work, the major record labels don’t want tracks to leak outside YouTube’s ecosystem. For this reason, they want YouTube to MP3 rippers shut down.

Earlier this month, the RIAA sent its overview of “notorious markets” to the Office of the US Trade Representative (USTR), highlighting several of these sites and asking for help.

“The overall popularity of these sites and the staggering volume of traffic it attracts evidences the enormous damage being inflicted on the U.S. record industry,” the RIAA wrote, calling out Mp3juices.cc, Convert2mp3.net, Savefrom.net, Ytmp3.cc, Convertmp3.io, Flvto.biz, and 2conv.com as the most popular offenders.

This position is shared by many other music industry groups. They see stream ripping as the largest piracy threat online. After shutting down YouTube-MP3, they hope to topple other sites as well, ideally with the backing of the US Government.

However, not everyone shares the belief that stream ripping equals copyright infringement.

In a rebuttal, the Electronic Frontier Foundation (EFF) informs the USTR that the RIAA is trying to twist the law in its favor. Not all stream ripping sites are facilitating copyright infringement by definition, the EFF argues.

“RIAA’s discussion of ‘stream-ripping’ websites misstates copyright law. Websites that simply allow users to extract the audio track from a user-selected online video are not ‘illegal sites’ and are not liable for copyright infringement, unless they engage in additional conduct that meets the definition of infringement,” the EFF writes.


While some people may use these sites to ‘pirate’ tracks there are also legitimate purposes, the digital rights group notes. Some creators specifically allow others to download and modify their work, for example, and in other cases ripping can be seen as fair use.

“There exists a vast and growing volume of online video that is licensed for free downloading and modification, or contains audio tracks that are not subject to copyright,” the EFF stresses.

“Moreover, many audio extractions qualify as non-infringing fair uses under copyright. Providing a service that is capable of extracting audio tracks for these lawful purposes is itself lawful, even if some users infringe.”

The fact that these sites generate revenue from advertising doesn’t make them illegal either. While there are some issues that could make a site liable, such as distributing infringing content to third parties, the EFF argues that many of the sites identified by the RIAA are not clearly involved in such activities.

Instead of solely relying on the characterizations of the RIAA, the US Government should judge these sites independently, in accordance with the law.

“USTR must apply U.S. law as it is, not as particular industry organizations wish it to be. Accordingly, it is inappropriate to describe ‘stream-ripping’ sites as engaging in or facilitating infringement. That logic would discourage U.S. firms from providing many forms of useful, lawful technology that processes or interacts with copyrighted work in digital form, to the detriment of U.S. trade,” the EFF concludes.

It is worth highlighting that most sites the RIAA mentioned specifically advertise themselves as YouTube converters. While this violates YouTube’s Terms of Service, something the streaming platform isn’t happy with, it doesn’t automatically classify them as infringing services.

Ideally, the RIAA and other music industry group would like YouTube to shut down these sites but if that doesn’t happen, more lawsuits may follow in the future. Then, the claims from both sides can be properly tested in court.

The full EFF response is available here (pdf). In addition to the stream ripping comments, the digital rights group also defends CDN providers such as Cloudflare, reverse proxies, and domain registrars from MPAA and RIAA piracy complaints.

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

OSSIM Download – Open Source SIEM Tools & Software

Post Syndicated from Darknet original https://www.darknet.org.uk/2017/10/ossim-download-open-source-siem-tools-software/?utm_source=rss&utm_medium=social&utm_campaign=darknetfeed

OSSIM Download – Open Source SIEM Tools & Software

OSSIM is a popular Open Source SIEM or Security Information and Event Management (SIEM) product, providing event collection, normalization and correlation.

OSSIM stands for Open Source Security Information Management, it was launched in 2003 by security engineers because of the lack of available open source products, OSSIM was created specifically to address the reality many security professionals face: A SIEM, whether it is open source or commercial, is virtually useless without the basic security controls necessary for security visibility.

Read the rest of OSSIM Download – Open Source SIEM Tools & Software now! Only available at Darknet.

timeShift(GrafanaBuzz, 1w) Issue 18

Post Syndicated from Blogs on Grafana Labs Blog original https://grafana.com/blog/2017/10/20/timeshiftgrafanabuzz-1w-issue-18/

Welcome to another issue of timeShift. This week we released Grafana 4.6.0-beta2, which includes some fixes for alerts, annotations, the Cloudwatch data source, and a few panel updates. We’re also gearing up for Oredev, one of the biggest tech conferences in Scandinavia, November 7-10. In addition to sponsoring, our very own Carl Bergquist will be presenting “Monitoring for everyone.” Hope to see you there – swing by our booth and say hi!

Latest Release

Grafana 4.6-beta-2 is now available! Grafana 4.6.0-beta2 adds fixes for:

  • ColorPicker display
  • Alerting test
  • Cloudwatch improvements
  • CSV export
  • Text panel enhancements
  • Annotation fix for MySQL

To see more details on what’s in the newest version, please see the release notes.

Download Grafana 4.6.0-beta-2 Now

From the Blogosphere

Screeps and Grafana: Graphing your AI: If you’re unfamiliar with Screeps, it’s a MMO RTS game for programmers, where the objective is to grow your colony through programming your units’ AI. You control your colony by writing JavaScript, which operates 247 in the single persistent real-time world filled by other players. This article walks you through graphing all your game stats with Grafana.

ntopng Grafana Integration: The Beauty of Data Visualization: Our friends at ntop created a tutorial so that you can graph ntop monitoring data in Grafana. He goes through the metrics exposed, configuring the ntopng Data Source plugin, and building your first dashboard. They’ve also created a nice video tutorial of the process.

Installing Graphite and Grafana to Display the Graphs of Centreon: This article, provides a step-by-step guide to getting your Centreon data into Graphite and visualizing the data in Grafana.

Bit v. Byte Episode 3 – Metrics for the Win: Bit v. Byte is a new weekly Podcast about the web industry, tools and techniques upcoming and in use today. This episode dives into metrics, and discusses Grafana, Prometheus and NGINX Amplify.

Code-Quickie: Visualize heating with Grafana: With the winter weather coming, Reinhard wanted to monitor the stats in his boiler room. This article covers not only the visualization of the data, but the different devices and sensors you can use to can use in your own home.

RuuviTag with C.H.I.P – BLE – Node-RED: Following the temperature-monitoring theme from the last article, Tobias writes about his journey of hooking up his new RuuviTag to Grafana to measure temperature, relative humidity, air pressure and more.

Early Bird will be Ending Soon

Early bird discounts will be ending soon, but you still have a few days to lock in the lower price. We will be closing early bird on October 31, so don’t wait until the last minute to take advantage of the discounted tickets!

Also, there’s still time to submit your talk. We’ll accept submissions through the end of October. We’re looking for technical and non-technical talks of all sizes. Submit a CFP now.

Get Your Early Bird Ticket Now

Grafana Plugins

This week we have updates to two panels and a brand new panel that can add some animation to your dashboards. Installing plugins in Grafana is easy; for on-prem Grafana, use the Grafana-cli tool, or with 1 click if you are using Hosted Grafana.


Geoloop Panel – The Geoloop panel is a simple visualizer for joining GeoJSON to Time Series data, and animating the geo features in a loop. An example of using the panel would be showing the rate of rainfall during a 5-hour storm.

Install Now


Breadcrumb Panel – This plugin keeps track of dashboards you have visited within one session and displays them as a breadcrumb. The latest update fixes some issues with back navigation and url query params.



Influx Admin Panel – The Influx Admin panel duplicates features from the now deprecated Web Admin Interface for InfluxDB and has lots of features like letting you see the currently running queries, which can also be easily killed.

Changes in the latest release:

  • Converted to typescript project based on typescript-template-datasource
  • Select Databases. This only works with PR#8096
  • Added time format options
  • Show tags from response
  • Support template variables in the query


Contribution of the week:

Each week we highlight some of the important contributions from our amazing open source community. Thank you for helping make Grafana better!

The Stockholm Go Meetup had a hackathon this week and sent a PR for letting whitelisted cookies pass through the Grafana proxy. Thanks to everyone who worked on this PR!

Tweet of the Week

We scour Twitter each week to find an interesting/beautiful dashboard and show it off! #monitoringLove

This is awesome – we can’t get enough of these public dashboards!

We Need Your Help!

Do you have a graph that you love because the data is beautiful or because the graph provides interesting information? Please get in touch. Tweet or send us an email with a screenshot, and we’ll tell you about this fun experiment.

Tell Me More

Grafana Labs is Hiring!

We are passionate about open source software and thrive on tackling complex challenges to build the future. We ship code from every corner of the globe and love working with the community. If this sounds exciting, you’re in luck – WE’RE HIRING!

Check out our Open Positions

How are we doing?

Please tell us how we’re doing. Submit a comment on this article below, or post something at our community forum. Help us make these weekly roundups better!

Follow us on Twitter, like us on Facebook, and join the Grafana Labs community.

Epic Games Sues Man Over Bitcoin Mining Fortnite ‘Cheat’

Post Syndicated from Ernesto original https://torrentfreak.com/epic-games-sues-man-over-bitcoin-mining-fortnite-cheat-171019/

A few weeks ago, Epic Games released Fortnite’s free-to-play “Battle Royale” game mode for the PC and other platforms, generating massive interest among gamers.

The release also attracted attention from thousands of cheaters, many of whom were subsequently banned. In addition, Epic Games went a step further by taking several cheaters to court over copyright infringement.

This week the North Carolina-based game developer continued its a war against cheaters. In a new lawsuit, it targets two other cheaters who promoted their hacks through YouTube videos.

One of the defendants is a Swedish resident, Mr. Josefson. He created a cheat and promoted it in various videos, adding instructions on how to download and install it. In common with the previous defendants, he is being sued for copyright infringement.

The second cheater listed in the complaint, a Russian man named Mr. Yakovenko, is more unique. This man also promoted his Fortnite cheats through a series of YouTube videos, but they weren’t very effective.

When Epic downloaded the ‘cheat’ to see how it works, all they got was a Bitcoin miner.

“Epic downloaded the purported cheat from the links provided in Yakovenko’s YouTube videos. While the ‘cheat’ does not appear to be a functional Fortnite cheat, it functions as a bitcoin miner that infects the user’s computer with a virus that causes the user’s computer to mine bitcoin for the benefit of an unknown third party,” the complaint reads.

Epic ‘cheat’

Despite the non-working cheat, Epic Games maintains that Yakovenko created a cheat for Fortnite’s Battle Royale game mode, pointing to a YouTube video he posted last month.

“The First Yakovenko video and associated post contained instructions on how to download and install the cheat and showed full screen gameplay using the purported cheat,” the complaint reads.

All the videos have since been removed following takedown notices from Epic. Through the lawsuit, the game developer now hopes to get compensation for the damages it suffered.

In addition to the copyright infringement claims the two men are also accused of trademark infringement, unfair competition, and breach of contract.

There’s little doubt that Epic Games is doing its best to hold cheaters accountable. However, the problem is not easy to contain. A simple search for Fortnite Hack or Fortnite Cheat still yields tens of thousands of results, with new videos being added continuously.

A copy of the full complaint against Josefson and Yakovenko is available here (pdf).

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

Federate Database User Authentication Easily with IAM and Amazon Redshift

Post Syndicated from Thiyagarajan Arumugam original https://aws.amazon.com/blogs/big-data/federate-database-user-authentication-easily-with-iam-and-amazon-redshift/

Managing database users though federation allows you to manage authentication and authorization procedures centrally. Amazon Redshift now supports database authentication with IAM, enabling user authentication though enterprise federation. No need to manage separate database users and passwords to further ease the database administration. You can now manage users outside of AWS and authenticate them for access to an Amazon Redshift data warehouse. Do this by integrating IAM authentication and a third-party SAML-2.0 identity provider (IdP), such as AD FS, PingFederate, or Okta. In addition, database users can also be automatically created at their first login based on corporate permissions.

In this post, I demonstrate how you can extend the federation to enable single sign-on (SSO) to the Amazon Redshift data warehouse.

SAML and Amazon Redshift

AWS supports Security Assertion Markup Language (SAML) 2.0, which is an open standard for identity federation used by many IdPs. SAML enables federated SSO, which enables your users to sign in to the AWS Management Console. Users can also make programmatic calls to AWS API actions by using assertions from a SAML-compliant IdP. For example, if you use Microsoft Active Directory for corporate directories, you may be familiar with how Active Directory and AD FS work together to enable federation. For more information, see the Enabling Federation to AWS Using Windows Active Directory, AD FS, and SAML 2.0 AWS Security Blog post.

Amazon Redshift now provides the GetClusterCredentials API operation that allows you to generate temporary database user credentials for authentication. You can set up an IAM permissions policy that generates these credentials for connecting to Amazon Redshift. Extending the IAM authentication, you can configure the federation of AWS access though a SAML 2.0–compliant IdP. An IAM role can be configured to permit the federated users call the GetClusterCredentials action and generate temporary credentials to log in to Amazon Redshift databases. You can also set up policies to restrict access to Amazon Redshift clusters, databases, database user names, and user group.

Amazon Redshift federation workflow

In this post, I demonstrate how you can use a JDBC– or ODBC-based SQL client to log in to the Amazon Redshift cluster using this feature. The SQL clients used with Amazon Redshift JDBC or ODBC drivers automatically manage the process of calling the GetClusterCredentials action, retrieving the database user credentials, and establishing a connection to your Amazon Redshift database. You can also use your database application to programmatically call the GetClusterCredentials action, retrieve database user credentials, and connect to the database. I demonstrate these features using an example company to show how different database users accounts can be managed easily using federation.

The following diagram shows how the SSO process works:

  2. Authenticate using Corp Username/Password
  3. IdP sends SAML assertion
  4. Call STS to assume role with SAML
  5. STS Returns Temp Credentials
  6. Use Temp Credentials to get Temp cluster credentials
  7. Connect to Amazon Redshift using temp credentials


Example Corp. is using Active Directory (idp host:demo.examplecorp.com) to manage federated access for users in its organization. It has an AWS account: 123456789012 and currently manages an Amazon Redshift cluster with the cluster ID “examplecorp-dw”, database “analytics” in us-west-2 region for its Sales and Data Science teams. It wants the following access:

  • Sales users can access the examplecorp-dw cluster using the sales_grp database group
  • Sales users access examplecorp-dw through a JDBC-based SQL client
  • Sales users access examplecorp-dw through an ODBC connection, for their reporting tools
  • Data Science users access the examplecorp-dw cluster using the data_science_grp database group.
  • Partners access the examplecorp-dw cluster and query using the partner_grp database group.
  • Partners are not federated through Active Directory and are provided with separate IAM user credentials (with IAM user name examplecorpsalespartner).
  • Partners can connect to the examplecorp-dw cluster programmatically, using language such as Python.
  • All users are automatically created in Amazon Redshift when they log in for the first time.
  • (Optional) Internal users do not specify database user or group information in their connection string. It is automatically assigned.
  • Data warehouse users can use SSO for the Amazon Redshift data warehouse using the preceding permissions.

Step 1:  Set up IdPs and federation

The Enabling Federation to AWS Using Windows Active Directory post demonstrated how to prepare Active Directory and enable federation to AWS. Using those instructions, you can establish trust between your AWS account and the IdP and enable user access to AWS using SSO.  For more information, see Identity Providers and Federation.

For this walkthrough, assume that this company has already configured SSO to their AWS account: 123456789012 for their Active Directory domain demo.examplecorp.com. The Sales and Data Science teams are not required to specify database user and group information in the connection string. The connection string can be configured by adding SAML Attribute elements to your IdP. Configuring these optional attributes enables internal users to conveniently avoid providing the DbUser and DbGroup parameters when they log in to Amazon Redshift.

The user-name attribute can be set up as follows, with a user ID (for example, nancy) or an email address (for example. [email protected]):

<Attribute Name="https://redshift.amazon.com/SAML/Attributes/DbUser">  

The AutoCreate attribute can be defined as follows:

<Attribute Name="https://redshift.amazon.com/SAML/Attributes/AutoCreate">

The sales_grp database group can be included as follows:

<Attribute Name="https://redshift.amazon.com/SAML/Attributes/DbGroups">

For more information about attribute element configuration, see Configure SAML Assertions for Your IdP.

Step 2: Create IAM roles for access to the Amazon Redshift cluster

The next step is to create IAM policies with permissions to call GetClusterCredentials and provide authorization for Amazon Redshift resources. To grant a SQL client the ability to retrieve the cluster endpoint, region, and port automatically, include the redshift:DescribeClusters action with the Amazon Redshift cluster resource in the IAM role.  For example, users can connect to the Amazon Redshift cluster using a JDBC URL without the need to hardcode the Amazon Redshift endpoint:

Previous:  jdbc:redshift://endpoint:port/database

Current:  jdbc:redshift:iam://clustername:region/dbname

Use IAM to create the following policies. You can also use an existing user or role and assign these policies. For example, if you already created an IAM role for IdP access, you can attach the necessary policies to that role. Here is the policy created for sales users for this example:


    "Version": "2012-10-17",
    "Statement": [
            "Effect": "Allow",
            "Action": [
            "Resource": [
            "Effect": "Allow",
            "Action": [
            "Resource": [
            "Condition": {
                "StringEquals": {
                    "aws:userid": "AIDIODR4TAW7CSEXAMPLE:${redshift:DbUser}@examplecorp.com"
            "Effect": "Allow",
            "Action": [
            "Resource": [
            "Effect": "Allow",
            "Action": [
            "Resource": [

The policy uses the following parameter values:

  • Region: us-west-2
  • AWS Account: 123456789012
  • Cluster name: examplecorp-dw
  • Database group: sales_grp
Policy Statement Description

Allow users to retrieve the cluster endpoint, region, and port automatically for the Amazon Redshift cluster examplecorp-dw. This specification uses the resource format arn:aws:redshift:region:account-id:cluster:clustername. For example, the SQL client JDBC can be specified in the format jdbc:redshift:iam://clustername:region/dbname.

For more information, see Amazon Resource Names.


Generates a temporary token to authenticate into the examplecorp-dw cluster. “arn:aws:redshift:us-west-2:123456789012:dbuser:examplecorp-dw/${redshift:DbUser}” restricts the corporate user name to the database user name for that user. This resource is specified using the format: arn:aws:redshift:region:account-id:dbuser:clustername/dbusername.

The Condition block enforces that the AWS user ID should match “AIDIODR4TAW7CSEXAMPLE:${redshift:DbUser}@examplecorp.com”, so that individual users can authenticate only as themselves. The AIDIODR4TAW7CSEXAMPLE role has the Sales_DW_IAM_Policy policy attached.

Automatically creates database users in examplecorp-dw, when they log in for the first time. Subsequent logins reuse the existing database user.
Allows sales users to join the sales_grp database group through the resource “arn:aws:redshift:us-west-2:123456789012:dbgroup:examplecorp-dw/sales_grp” that is specified in the format arn:aws:redshift:region:account-id:dbgroup:clustername/dbgroupname.

Similar policies can be created for Data Science users with access to join the data_science_grp group in examplecorp-dw. You can now attach the Sales_DW_IAM_Policy policy to the role that is mapped to IdP application for SSO.
 For more information about how to define the claim rules, see Configuring SAML Assertions for the Authentication Response.

Because partners are not authorized using Active Directory, they are provided with IAM credentials and added to the partner_grp database group. The Partner_DW_IAM_Policy is attached to the IAM users for partners. The following policy allows partners to log in using the IAM user name as the database user name.


    "Version": "2012-10-17",
    "Statement": [
            "Effect": "Allow",
            "Action": [
            "Resource": [
            "Effect": "Allow",
            "Action": [
            "Resource": [
            "Condition": {
                "StringEquals": {
                    "redshift:DbUser": "${aws:username}"
            "Effect": "Allow",
            "Action": [
            "Resource": [
            "Effect": "Allow",
            "Action": [
            "Resource": [

redshift:DbUser“: “${aws:username}” forces an IAM user to use the IAM user name as the database user name.

With the previous steps configured, you can now establish the connection to Amazon Redshift through JDBC– or ODBC-supported clients.

Step 3: Set up database user access

Before you start connecting to Amazon Redshift using the SQL client, set up the database groups for appropriate data access. Log in to your Amazon Redshift database as superuser to create a database group, using CREATE GROUP.

Log in to examplecorp-dw/analytics as superuser and create the following groups and users:

CREATE GROUP sales_grp;
CREATE GROUP datascience_grp;
CREATE GROUP partner_grp;

Use the GRANT command to define access permissions to database objects (tables/views) for the preceding groups.

Step 4: Connect to Amazon Redshift using the JDBC SQL client

Assume that sales user “nancy” is using the SQL Workbench client and JDBC driver to log in to the Amazon Redshift data warehouse. The following steps help set up the client and establish the connection:

  1. Download the latest Amazon Redshift JDBC driver from the Configure a JDBC Connection page
  2. Build the JDBC URL with the IAM option in the following format:

Because the redshift:DescribeClusters action is assigned to the preceding IAM roles, it automatically resolves the cluster endpoints and the port. Otherwise, you can specify the endpoint and port information in the JDBC URL, as described in Configure a JDBC Connection.

Identify the following JDBC options for providing the IAM credentials (see the “Prepare your environment” section) and configure in the SQL Workbench Connection Profile:

idp_host=demo.examplecorp.com (The name of the corporate identity provider host)
idp_port=443  (The port of the corporate identity provider host)
user=examplecorp\nancy(corporate user name)
password=***(corporate user password)

The SQL workbench configuration looks similar to the following screenshot:

Now, “nancy” can connect to examplecorp-dw by authenticating using the corporate Active Directory. Because the SAML attributes elements are already configured for nancy, she logs in as database user nancy and is assigned the sales_grp. Similarly, other Sales and Data Science users can connect to the examplecorp-dw cluster. A custom Amazon Redshift ODBC driver can also be used to connect using a SQL client. For more information, see Configure an ODBC Connection.

Step 5: Connecting to Amazon Redshift using JDBC SQL Client and IAM Credentials

This optional step is necessary only when you want to enable users that are not authenticated with Active Directory. Partners are provided with IAM credentials that they can use to connect to the examplecorp-dw Amazon Redshift clusters. These IAM users are attached to Partner_DW_IAM_Policy that assigns them to be assigned to the public database group in Amazon Redshift. The following JDBC URLs enable them to connect to the Amazon Redshift cluster:

jdbc:redshift:iam//examplecorp-dw/analytics?AccessKeyID=XXX&SecretAccessKey=YYY&DbUser=examplecorpsalespartner&DbGroup= partner_grp&AutoCreate=true

The AutoCreate option automatically creates a new database user the first time the partner logs in. There are several other options available to conveniently specify the IAM user credentials. For more information, see Options for providing IAM credentials.

Step 6: Connecting to Amazon Redshift using an ODBC client for Microsoft Windows

Assume that another sales user “uma” is using an ODBC-based client to log in to the Amazon Redshift data warehouse using Example Corp Active Directory. The following steps help set up the ODBC client and establish the Amazon Redshift connection in a Microsoft Windows operating system connected to your corporate network:

  1. Download and install the latest Amazon Redshift ODBC driver.
  2. Create a system DSN entry.
    1. In the Start menu, locate the driver folder or folders:
      • Amazon Redshift ODBC Driver (32-bit)
      • Amazon Redshift ODBC Driver (64-bit)
      • If you installed both drivers, you have a folder for each driver.
    2. Choose ODBC Administrator, and then type your administrator credentials.
    3. To configure the driver for all users on the computer, choose System DSN. To configure the driver for your user account only, choose User DSN.
    4. Choose Add.
  3. Select the Amazon Redshift ODBC driver, and choose Finish. Configure the following attributes:
    Data Source Name =any friendly name to identify the ODBC connection 
    user=uma(corporate user name)
    Auth Type-Identity Provider: AD FS
    password=leave blank (Windows automatically authenticates)
    Cluster ID: examplecorp-dw
    idp_host=demo.examplecorp.com (The name of the corporate IdP host)

This configuration looks like the following:

  1. Choose OK to save the ODBC connection.
  2. Verify that uma is set up with the SAML attributes, as described in the “Set up IdPs and federation” section.

The user uma can now use this ODBC connection to establish the connection to the Amazon Redshift cluster using any ODBC-based tools or reporting tools such as Tableau. Internally, uma authenticates using the Sales_DW_IAM_Policy  IAM role and is assigned the sales_grp database group.

Step 7: Connecting to Amazon Redshift using Python and IAM credentials

To enable partners, connect to the examplecorp-dw cluster programmatically, using Python on a computer such as Amazon EC2 instance. Reuse the IAM users that are attached to the Partner_DW_IAM_Policy policy defined in Step 2.

The following steps show this set up on an EC2 instance:

  1. Launch a new EC2 instance with the Partner_DW_IAM_Policy role, as described in Using an IAM Role to Grant Permissions to Applications Running on Amazon EC2 Instances. Alternatively, you can attach an existing IAM role to an EC2 instance.
  2. This example uses Python PostgreSQL Driver (PyGreSQL) to connect to your Amazon Redshift clusters. To install PyGreSQL on Amazon Linux, use the following command as the ec2-user:
    sudo easy_install pip
    sudo yum install postgresql postgresql-devel gcc python-devel
    sudo pip install PyGreSQL

  1. The following code snippet demonstrates programmatic access to Amazon Redshift for partner users:
    #!/usr/bin/env python
    python redshift-unload-copy.py <config file> <region>
    * Copyright 2014, Amazon.com, Inc. or its affiliates. All Rights Reserved.
    * Licensed under the Amazon Software License (the "License").
    * You may not use this file except in compliance with the License.
    * A copy of the License is located at
    * http://aws.amazon.com/asl/
    * or in the "license" file accompanying this file. This file is distributed
    * express or implied. See the License for the specific language governing
    * permissions and limitations under the License.
    import sys
    import pg
    import boto3
    REGION = 'us-west-2'
    CLUSTER_IDENTIFIER = 'examplecorp-dw'
    DB_NAME = 'sales_db'
    DB_USER = 'examplecorpsalespartner'
    options = """keepalives=1 keepalives_idle=200 keepalives_interval=200
    set_timeout_stmt = "set statement_timeout = 1200000"
    def conn_to_rs(host, port, db, usr, pwd, opt=options, timeout=set_timeout_stmt):
        rs_conn_string = """host=%s port=%s dbname=%s user=%s password=%s
                             %s""" % (host, port, db, usr, pwd, opt)
        print "Connecting to %s:%s:%s as %s" % (host, port, db, usr)
        rs_conn = pg.connect(dbname=rs_conn_string)
        return rs_conn
    def main():
        # describe the cluster and fetch the IAM temporary credentials
        global redshift_client
        redshift_client = boto3.client('redshift', region_name=REGION)
        response_cluster_details = redshift_client.describe_clusters(ClusterIdentifier=CLUSTER_IDENTIFIER)
        response_credentials = redshift_client.get_cluster_credentials(DbUser=DB_USER,DbName=DB_NAME,ClusterIdentifier=CLUSTER_IDENTIFIER,DurationSeconds=3600)
        rs_host = response_cluster_details['Clusters'][0]['Endpoint']['Address']
        rs_port = response_cluster_details['Clusters'][0]['Endpoint']['Port']
        rs_db = DB_NAME
        rs_iam_user = response_credentials['DbUser']
        rs_iam_pwd = response_credentials['DbPassword']
        # connect to the Amazon Redshift cluster
        conn = conn_to_rs(rs_host, rs_port, rs_db, rs_iam_user,rs_iam_pwd)
        # execute a query
        result = conn.query("SELECT sysdate as dt")
        # fetch results from the query
        for dt_val in result.getresult() :
            print dt_val
        # close the Amazon Redshift connection
    if __name__ == "__main__":

You can save this Python program in a file (redshiftscript.py) and execute it at the command line as ec2-user:

python redshiftscript.py

Now partners can connect to the Amazon Redshift cluster using the Python script, and authentication is federated through the IAM user.


In this post, I demonstrated how to use federated access using Active Directory and IAM roles to enable single sign-on to an Amazon Redshift cluster. I also showed how partners outside an organization can be managed easily using IAM credentials.  Using the GetClusterCredentials API action, now supported by Amazon Redshift, lets you manage a large number of database users and have them use corporate credentials to log in. You don’t have to maintain separate database user accounts.

Although this post demonstrated the integration of IAM with AD FS and Active Directory, you can replicate this solution across with your choice of SAML 2.0 third-party identity providers (IdP), such as PingFederate or Okta. For the different supported federation options, see Configure SAML Assertions for Your IdP.

If you have questions or suggestions, please comment below.

Additional Reading

Learn how to establish federated access to your AWS resources by using Active Directory user attributes.

About the Author

Thiyagarajan Arumugam is a Big Data Solutions Architect at Amazon Web Services and designs customer architectures to process data at scale. Prior to AWS, he built data warehouse solutions at Amazon.com. In his free time, he enjoys all outdoor sports and practices the Indian classical drum mridangam.


Google Asked to Remove 3 Billion “Pirate” Search Results

Post Syndicated from Ernesto original https://torrentfreak.com/google-asked-to-remove-3-billion-pirate-search-results-171018/

Copyright holders continue to flood Google with DMCA takedown requests, asking the company to remove “pirate links” from its search results.

In recent years the number of reported URLs has exploded, surging to unprecedented heights.

Since Google first started to report the volume of takedown requests in its Transparency Report, the company has been asked to remove more than three billion allegedly infringing search results.

The frequency at which these URLs are reported has increased over the years and at the moment roughly three million ‘pirate’ URLs are submitted per day.

The URLs are sent in by major rightsholders including members of the BPI, RIAA, and various major Hollywood studios. They target a wide variety of sites, over 1.3 million, but a few dozen ‘repeat offenders’ are causing the most trouble.

File-hosting service 4shared.com currently tops the list of most-targeted domains with 66 million URLs, followed by the now-defunct MP3 download site MP3toys.xyz and Rapidgator.net, with 51 and 28 million URLs respectively.

3 billion URLs

Interestingly, the high volume of takedown notices is used as an argument for and against the DMCA process.

While Google believes that the millions of reported URLs per day are a sign that the DMCA takedown process is working correctly, rightsholders believe the volumes are indicative of an unbeatable game of whack-a-mole.

According to some copyright holders, the takedown efforts do little to seriously combat piracy. Various industry groups have therefore asked governments and lawmakers for broad revisions.

Among other things they want advanced technologies and processes to ensure that infringing content doesn’t reappear elsewhere once it’s removed, a so-called “notice and stay down” approach. In addition, Google has often been asked to demote pirate links in search results.

UK music industry group BPI, who are responsible for more than 10% of all the takedown requests on Google, sees the new milestone as an indicator of how much effort its anti-piracy activities take.

“This 3 billion figure shows how hard the creative sector has to work to police its content online and how much time and resource this takes. The BPI is the world’s largest remover of illegal music links from Google, one third of which are on behalf of independent record labels,” Geoff Taylor, BPI’s Chief Executive, informs TF.

However, there is also some progress to report. Earlier this year BPI announced a voluntary partnership with Google and Bing to demote pirate content faster and more effectively for US visitors.

“We now have a voluntary code of practice in place in the UK, facilitated by Government, that requires Google and Bing to work together with the BPI and other creator organizations to develop lasting solutions to the problem of illegal sites gaining popularity in search listings,” Taylor notes.

According to BPI, both Google and Bing have shown that changes to their algorithms can be effective in demoting the worst pirate sites from the top search results and they hope others will follow suit.

“Other intermediaries should follow this lead and take more responsibility to work with creators to reduce the proliferation of illegal links and disrupt the ability of illegal sites to capture consumers and build black market businesses that take money away from creators.”

Agreement or not, there are still plenty of pirate links in search results, so the BPI is still sending out millions of takedown requests per month.

We asked Google for a comment on the new milestone but at the time of writing, we have yet to hear back. In any event, the issue is bound to remain a hot topic during the months and years to come.

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

What’s new in HiveMQ 3.3

Post Syndicated from The HiveMQ Team original https://www.hivemq.com/whats-new-in-hivemq-3-3

We are pleased to announce the release of HiveMQ 3.3. This version of HiveMQ is the most advanced and user friendly version of HiveMQ ever. A broker is the heart of every MQTT deployment and it’s key to monitor and understand how healthy your system and your connected clients are. Version 3.3 of HiveMQ focuses on observability, usability and advanced administration features and introduces a brand new Web UI. This version is a drop-in replacement for HiveMQ 3.2 and of course supports rolling upgrades for zero-downtime.

HiveMQ 3.3 brings many features that your users, administrators and plugin developers are going to love. These are the highlights:

Web UI

Web UI
The new HiveMQ version has a built-in Web UI for advanced analysis and administrative tasks. A powerful dashboard shows important data about the health of the broker cluster and an overview of the whole MQTT deployment.
With the new Web UI, administrators are able to drill down to specific client information and can perform administrative actions like disconnecting a client. Advanced analytics functionality allows indetifying clients with irregular behavior. It’s easy to identify message-dropping clients as HiveMQ shows detailed statistics of such misbehaving MQTT participants.
Of course all Web UI features work at scale with more than a million connected MQTT clients. Learn more about the Web UI in the documentation.

Time To Live

HiveMQ introduces Time to Live (TTL) on various levels of the MQTT lifecycle. Automatic cleanup of expired messages is as well supported as the wiping of abandoned persistent MQTT sessions. In particular, version 3.3 implements the following TTL features:

  • MQTT client session expiration
  • Retained Message expiration
  • MQTT PUBLISH message expiration

Configuring a TTL for MQTT client sessions and retained messages allows freeing system resources without manual administrative intervention as soon as the data is not needed anymore.
Beside global configuration, MQTT PUBLISHES can have individual TTLs based on application specific characteristics. It’s a breeze to change the TTL of particular messages with the HiveMQ plugin system. As soon as a message TTL expires, the broker won’t send out the message anymore, even if the message was previously queued or in-flight. This can save precious bandwidth for mobile connections as unnecessary traffic is avoided for expired messages.

Trace Recordings

Trace Recordings
Debugging specific MQTT clients or groups of MQTT clients can be challenging at scale. HiveMQ 3.3 introduces an innovative Trace Recording mechanism that allows creating detailed recordings of all client interactions with given filters.
It’s possible to filter based on client identifiers, MQTT message types and topics. And the best of all: You can use regular expressions to select multiple MQTT clients at once as well as topics with complex structures. Getting detailed information about the behavior of specific MQTT clients for debugging complex issues was never easier.

Native SSL

Native SSL
The new native SSL integration of HiveMQ brings a performance boost of more than 40% for SSL Handshakes (in terms of CPU usage) by utilizing an integration with BoringSSL. BoringSSL is Google’s fork of OpenSSL which is also used in Google Chrome and Android. Besides the compute and huge memory optimizations (saves up to 60% Java Heap), additional secure state-of-the-art cipher suites are supported by HiveMQ which are not directly available for Java (like ChaCha20-Poly1305).
Most HiveMQ deployments on Linux systems are expected to see decreased CPU load on TLS handshakes with the native SSL integration and huge memory improvements.

New Plugin System Features

New Plugin System Features
The popular and powerful plugin system has received additional services and callbacks which are useful for many existing and future plugins.
Plugin developers can now use a ConnectionAttributeStore and a SessionAttributeStore for storing arbitrary data for the lifetime of a single MQTT connection of a client or for the whole session of a client. The new ClientGroupService allows grouping different MQTT client identifiers by the same key, so it’s easy to address multiple MQTT clients (with the same group) at once.

A new callback was introduced which notifies a plugin when a HiveMQ instance is ready, which means the instance is part of the cluster and all listeners were started successfully. Developers can now react when a MQTT client session is ready and usable in the cluster with a dedicated callback.

Some use cases require modifying a MQTT PUBLISH packet before it’s sent out to a client. This is now possible with a new callback that was introduced for modifying a PUBLISH before sending it out to a individual client.
The offline queue size for persistent clients is now also configurable for individual clients as well as the queue discard strategy.

Additional Features

Additional Features
HiveMQ 3.3 has many additional features designed for power users and professional MQTT deployments. The new version also has the following highlights:

  • OCSP Stapling
  • Event Log for MQTT client connects, disconnects and unusual events (e.g. discarded message due to slow consumption on the client side
  • Throttling of concurrent TLS handshakes
  • Connect Packet overload protection
  • Configuration of Socket send and receive buffer sizes
  • Global System Information like the HiveMQ Home folder can now be set via Environment Variables without changing the run script
  • The internal HTTP server of HiveMQ is now exposed to the holistic monitoring subsystem
  • Many additional useful metrics were exposed to HiveMQ’s monitoring subsystem


In order to upgrade to HiveMQ 3.3 from HiveMQ 3.2 or older versions, take a look at our Upgrade Guide.
Don’t forget to learn more about all the new features with our HiveMQ User Guide.

Download HiveMQ 3.3 now

Google Asked to Delist Pirate Movie Sites, ISPs Asked to Block Them

Post Syndicated from Andy original https://torrentfreak.com/google-asked-to-delist-pirate-movie-sites-isps-asked-to-block-them-171018/

After seizing several servers operated by popular private music tracker What.cd, last November French police went after a much bigger target.

Boasting millions of regular visitors, Zone-Telechargement (Zone-Download) was ranked the 11th most-visited website in the whole of the country. The site offered direct downloads of a wide variety of pirated content, including films, series, games, and music. Until the French Gendarmerie shut it down, that is.

After being founded in 2011 and enjoying huge growth following the 2012 raids against Megaupload, the Zone-Telechargement ‘brand’ was still popular with French users, despite the closure of the platform. It, therefore, came as no surprise that the site was quickly cloned by an unknown party and relaunched as Zone-Telechargement.ws.

The site has been doing extremely well following its makeover. To the annoyance of copyright holders, SimilarWeb reports the platform as France’s 37th most popular site with around 58 million visitors per month. That’s a huge achievement in less than 12 months.

Now, however, the site is receiving more unwanted attention. PCInpact says it has received information that several movie-focused organizations including the French National Film Center are requesting tough action against the site.

The National Federation of Film Distributors, the Video Publishing Union, the Association of Independent Producers and the Producers Union are all demanding the blocking of Zone-Telechargement by several local ISPs, alongside its delisting from search results.

The publication mentions four Internet service providers – Free, Numericable, Bouygues Telecom, and Orange – plus Google on the search engine front. At this stage, other search companies, such as Microsoft’s Bing, are not reported as part of the action.

In addition to Zone-Telechargement, several other ‘pirate’ sites (Papystreaming.org, Sokrostream.cc and Zonetelechargement.su, another site playing on the popular brand) are included in the legal process. All are described as “structurally infringing” by the complaining movie outfits, PCInpact notes.

The legal proceedings against the sites are based in Article 336-2 of the Intellectual Property Code. It’s ground already trodden by movie companies who following a 2011 complaint, achieved victory in 2013 against several Allostreaming-linked sites.

In that case, the High Court of Paris ordered ISPs, several of which appear in the current action, to “implement all appropriate means including blocking” to prevent access to the infringing sites.

The Court also ordered Google, Microsoft, and Yahoo to “take all necessary measures to prevent the occurrence on their services of any results referring to any of the sites” on their platforms.

Also of interest is that the action targets a service called DL-Protecte.com, which according to local anti-piracy agency HADOPI, makes it difficult for rightsholders to locate infringing content while at the same time generates more revenue for pirate sites.

A judgment is expected in “several months.”

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

Spaghetti Download – Web Application Security Scanner

Post Syndicated from Darknet original https://www.darknet.org.uk/2017/10/spaghetti-download-web-application-security-scanner/?utm_source=rss&utm_medium=social&utm_campaign=darknetfeed

Spaghetti Download – Web Application Security Scanner

Spaghetti is an Open-source Web Application Security Scanner, it is designed to find various default and insecure files, configurations, and misconfigurations.

It is built on Python 2.7 and can run on any platform which has a Python environment.

Features of Spaghetti Web Application Security Scanner

  • Fingerprints
    • Server
    • Web Frameworks (CakePHP, CherryPy,…)
    • Web Application Firewall (Waf)
    • Content Management System (CMS)
    • Operating System (Linux, Unix,..)
    • Language (PHP, Ruby,…)
    • Cookie Security
  • Bruteforce
    • Admin Interface
    • Common Backdoors
    • Common Backup Directory
    • Common Backup File
    • Common Directory
    • Common File
    • Log File
  • Disclosure
    • Emails
    • Private IP
    • Credit Cards
  • Attacks
    • HTML Injection
    • SQL Injection
    • LDAP Injection
    • XPath Injection
    • Cross Site Scripting (XSS)
    • Remote File Inclusion (RFI)
    • PHP Code Injection
  • Other
    • HTTP Allow Methods
    • HTML Object
    • Multiple Index
    • Robots Paths
    • Web Dav
    • Cross Site Tracing (XST)
    • .Listing
  • Vulns
    • ShellShock
    • Anonymous Cipher (CVE-2007-1858)
    • Crime (SPDY) (CVE-2012-4929)
    • Struts-Shock

Using Spaghetti Web Application Security Scanner

[email protected]:~/Spaghetti# python spaghetti.py
_____ _ _ _ _
| __|___ ___ ___| |_ ___| |_| |_|_|
|__ | .

Read the rest of Spaghetti Download – Web Application Security Scanner now! Only available at Darknet.

PureVPN Explains How it Helped the FBI Catch a Cyberstalker

Post Syndicated from Andy original https://torrentfreak.com/purevpn-explains-how-it-helped-the-fbi-catch-a-cyberstalker-171016/

Early October, Ryan S. Lin, 24, of Newton, Massachusetts, was arrested on suspicion of conducting “an extensive cyberstalking campaign” against a 24-year-old Massachusetts woman, as well as her family members and friends.

The Department of Justice described Lin’s offenses as a “multi-faceted” computer hacking and cyberstalking campaign. Launched in April 2016 when he began hacking into the victim’s online accounts, Lin allegedly obtained personal photographs and sensitive information about her medical and sexual histories and distributed that information to hundreds of other people.

Details of what information the FBI compiled on Lin can be found in our earlier report but aside from his alleged crimes (which are both significant and repugnant), it was PureVPN’s involvement in the case that caused the most controversy.

In a report compiled by an FBI special agent, it was revealed that the Hong Kong-based company’s logs helped the authorities net the alleged criminal.

“Significantly, PureVPN was able to determine that their service was accessed by the same customer from two originating IP addresses: the RCN IP address from the home Lin was living in at the time, and the software company where Lin was employed at the time,” the agent’s affidavit reads.

Among many in the privacy community, this revelation was met with disappointment. On the PureVPN website the company claims to carry no logs and on a general basis, it’s expected that so-called “no-logging” VPN providers should provide people with some anonymity, at least as far as their service goes. Now, several days after the furor, the company has responded to its critics.

In a fairly lengthy statement, the company begins by confirming that it definitely doesn’t log what websites a user views or what content he or she downloads.

“PureVPN did not breach its Privacy Policy and certainly did not breach your trust. NO browsing logs, browsing habits or anything else was, or ever will be shared,” the company writes.

However, that’s only half the problem. While it doesn’t log user activity (what sites people visit or content they download), it does log the IP addresses that customers use to access the PureVPN service. These, given the right circumstances, can be matched to external activities thanks to logs carried by other web companies.

PureVPN talks about logs held by Google’s Gmail service to illustrate its point.

“A network log is automatically generated every time a user visits a website. For the sake of this example, let’s say a user logged into their Gmail account. Every time they accessed Gmail, the email provider created a network log,” the company explains.

“If you are using a VPN, Gmail’s network log would contain the IP provided by PureVPN. This is one half of the picture. Now, if someone asks Google who accessed the user’s account, Google would state that whoever was using this IP, accessed the account.

“If the user was connected to PureVPN, it would be a PureVPN IP. The inquirer [in the Lin case, the FBI] would then share timestamps and network logs acquired from Google and ask them to be compared with the network logs maintained by the VPN provider.”

Now, if PureVPN carried no logs – literally no logs – it would not be able to help with this kind of inquiry. That was the case last year when the FBI approached Private Internet Access for information and the company was unable to assist.

However, as is made pretty clear by PureVPN’s explanation, the company does log user IP addresses and timestamps which reveal when a user was logged on to the service. It doesn’t matter that PureVPN doesn’t log what the user allegedly did online, since the third-party service already knows that information to the precise second.

Following the example, GMail knows that a user sent an email at 10:22am on Monday October 16 from a PureVPN IP address. So, if PureVPN is approached by the FBI, the company can confirm that User X was using the same IP address at exactly the same time, and his home IP address was XXX.XX.XXX.XX. Effectively, the combined logs link one IP address to the other and the user is revealed. It’s that simple.

It is for this reason that in TorrentFreak’s annual summary of no-logging VPN providers, the very first question we ask every single company reads as follows:

Do you keep ANY logs which would allow you to match an IP-address and a time stamp to a user/users of your service? If so, what information do you hold and for how long?

Clearly, if a company says “yes we log incoming IP addresses and associated timestamps”, any claim to total user anonymity is ended right there and then.

While not completely useless (a logging service will still stop the prying eyes of ISPs and similar surveillance, while also defeating throttling and site-blocking), if you’re a whistle-blower with a job or even your life to protect, this level of protection is entirely inadequate.

The take-home points from this controversy are numerous, but perhaps the most important is for people to read and understand VPN provider logging policies.

Secondly, and just as importantly, VPN providers need to be extremely clear about the information they log. Not tracking browsing or downloading activities is all well and good, but if home IP addresses and timestamps are stored, this needs to be made clear to the customer.

Finally, VPN users should not be evil. There are plenty of good reasons to stay anonymous online but cyberstalking, death threats and ruining people’s lives are not included. Fortunately, the FBI have offline methods for catching this type of offender, and long may that continue.

PureVPN’s blog post is available here.

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

Some notes on the KRACK attack

Post Syndicated from Robert Graham original http://blog.erratasec.com/2017/10/some-notes-on-krack-attack.html

This is my interpretation of the KRACK attacks paper that describes a way of decrypting encrypted WiFi traffic with an active attack.

tl;dr: Wow. Everyone needs to be afraid. (Well, worried — not panicked.) It means in practice, attackers can decrypt a lot of wifi traffic, with varying levels of difficulty depending on your precise network setup. My post last July about the DEF CON network being safe was in error.


This is not a crypto bug but a protocol bug (a pretty obvious and trivial protocol bug).
When a client connects to the network, the access-point will at some point send a random “key” data to use for encryption. Because this packet may be lost in transmission, it can be repeated many times.
What the hacker does is just repeatedly sends this packet, potentially hours later. Each time it does so, it resets the “keystream” back to the starting conditions. The obvious patch that device vendors will make is to only accept the first such packet it receives, ignore all the duplicates.
At this point, the protocol bug becomes a crypto bug. We know how to break crypto when we have two keystreams from the same starting position. It’s not always reliable, but reliable enough that people need to be afraid.
Android, though, is the biggest danger. Rather than simply replaying the packet, a packet with key data of all zeroes can be sent. This allows attackers to setup a fake WiFi access-point and man-in-the-middle all traffic.
In a related case, the access-point/base-station can sometimes also be attacked, affecting the stream sent to the client.
Not only is sniffing possible, but in some limited cases, injection. This allows the traditional attack of adding bad code to the end of HTML pages in order to trick users into installing a virus.

This is an active attack, not a passive attack, so in theory, it’s detectable.

Who is vulnerable?

Everyone, pretty much.
The hacker only needs to be within range of your WiFi. Your neighbor’s teenage kid is going to be downloading and running the tool in order to eavesdrop on your packets.
The hacker doesn’t need to be logged into your network.
It affects all WPA1/WPA2, the personal one with passwords that we use in home, and the enterprise version with certificates we use in enterprises.
It can’t defeat SSL/TLS or VPNs. Thus, if you feel your laptop is safe surfing the public WiFi at airports, then your laptop is still safe from this attack. With Android, it does allow running tools like sslstrip, which can fool many users.
Your home network is vulnerable. Many devices will be using SSL/TLS, so are fine, like your Amazon echo, which you can continue to use without worrying about this attack. Other devices, like your Phillips lightbulbs, may not be so protected.

How can I defend myself?

More to the point, measure your current vendors by how long it takes them to patch. Throw away gear by those vendors that took a long time to patch and replace it with vendors that took a short time.
High-end access-points that contains “WIPS” (WiFi Intrusion Prevention Systems) features should be able to detect this and block vulnerable clients from connecting to the network (once the vendor upgrades the systems, of course). Even low-end access-points, like the $30 ones you get for home, can easily be updated to prevent packet sequence numbers from going back to the start (i.e. from the keystream resetting back to the start).
At some point, you’ll need to run the attack against yourself, to make sure all your devices are secure. Since you’ll be constantly allowing random phones to connect to your network, you’ll need to check their vulnerability status before connecting them. You’ll need to continue doing this for several years.
Of course, if you are using SSL/TLS for everything, then your danger is mitigated. This is yet another reason why you should be using SSL/TLS for internal communications.
Most security vendors will add things to their products/services to defend you. While valuable in some cases, it’s not a defense. The defense is patching the devices you know about, and preventing vulnerable devices from attaching to your network.
If I remember correctly, DEF CON uses Aruba. Aruba contains WIPS functionality, which means by the time DEF CON roles around again next year, they should have the feature to deny vulnerable devices from connecting, and specifically to detect an attack in progress and prevent further communication.
However, for an attacker near an Android device using a low-powered WiFi, it’s likely they will be able to conduct man-in-the-middle without any WIPS preventing them.

Top 10 Most Pirated Movies of The Week on BitTorrent – 10/16/17

Post Syndicated from Ernesto original https://torrentfreak.com/top-10-pirated-movies-week-bittorrent-101617/

This week we have two newcomers in our chart.

War for the Planet of the Apes is the most downloaded movie.

The data for our weekly download chart is estimated by TorrentFreak, and is for informational and educational reference only. All the movies in the list are Web-DL/Webrip/HDRip/BDrip/DVDrip unless stated otherwise.

RSS feed for the weekly movie download chart.

This week’s most downloaded movies are:
Movie Rank Rank last week Movie name IMDb Rating / Trailer
Most downloaded movies via torrents
1 (2) War for the Planet of the Apes 7.8 / trailer
2 (9) The Dark Tower 5.9 / trailer
3 (1) Spider-Man: Homecoming 7.8 / trailer
4 (…) American Made (Subbed HDrip) 7.3 / trailer
5 (3) Baby Driver 8.0 / trailer
6 (…) Annabelle Creation (Subbed HDRip) 6.7 / trailer
7 (7) Wonder Woman 8.2 / trailer
8 (4) Pirates of the Caribbean: Dead Men Tell No Tales 6.9 / trailer
9 (5) Transformers: The Last Knight 5.2 / trailer
10 (8) Despicable Me 3 6.4 / trailer

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

Netflix Expands Content Protection Team to Reduce Piracy

Post Syndicated from Ernesto original https://torrentfreak.com/netflix-expands-content-protection-team-to-reduce-piracy-171015/

There is little doubt that, in the United States and many other countries, Netflix has become the standard for watching movies on the Internet.

Despite the widespread availability, however, Netflix originals are widely pirated. Episodes from House of Cards, Narcos, and Orange is the New Black are downloaded and streamed millions of times through unauthorized platforms.

The streaming giant is obviously not happy with this situation and has ramped up its anti-piracy efforts in recent years. Since last year the company has sent out over a million takedown requests to Google alone and this volume continues to expand.

This growth coincides with an expansion of the company’s internal anti-piracy division. A new job posting shows that Netflix is expanding this team with a Copyright and Content Protection Coordinator. The ultimate goal is to reduce piracy to a fringe activity.

“The growing Global Copyright & Content Protection Group is looking to expand its team with the addition of a coordinator,” the job listing reads.

“He or she will be tasked with supporting the Netflix Global Copyright & Content Protection Group in its internal tactical take down efforts with the goal of reducing online piracy to a socially unacceptable fringe activity.”

Among other things, the new coordinator will evaluate new technological solutions to tackle piracy online.

More old-fashioned takedown efforts are also part of the job. This includes monitoring well-known content platforms, search engines and social network sites for pirated content.

“Day to day scanning of Facebook, YouTube, Twitter, Periscope, Google Search, Bing Search, VK, DailyMotion and all other platforms (including live platforms) used for piracy,” is listed as one of the main responsibilities.

Netflix’ Copyright and Content Protection Coordinator Job

The coordinator is further tasked with managing Facebook’s Rights Manager and YouTube’s Content-ID system, to prevent circumvention of these piracy filters. Experience with fingerprinting technologies and other anti-piracy tools will be helpful in this regard.

Netflix doesn’t do all the copyright enforcement on its own though. The company works together with other media giants in the recently launched “Alliance for Creativity and Entertainment” that is spearheaded by the MPAA.

In addition, the company also uses the takedown services of external anti-piracy outfits to target more traditional infringement sources, such as cyberlockers and piracy streaming sites. The coordinator has to keep an eye on these as well.

“Liaise with our vendors on manual takedown requests on linking sites and hosting sites and gathering data on pirate streaming sites, cyberlockers and usenet platforms.”

The above shows that Netflix is doing its best to prevent piracy from getting out of hand. It’s definitely taking the issue more seriously than a few years ago when the company didn’t have much original content.

The switch from being merely a distribution platform to becoming a major content producer and copyright holder has changed the stakes. Netflix hasn’t won the war on piracy, it’s just getting started.

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

‘Pirate’ EBook Site Refuses Point Blank to Cooperate With BREIN

Post Syndicated from Andy original https://torrentfreak.com/pirate-ebook-site-refuses-point-blank-to-cooperate-with-brein-171015/

Dutch anti-piracy group BREIN is probably best known for its legal action against The Pirate Bay but the outfit also tackles many other forms of piracy.

A prime example is the case it pursued against a seller of fully-loaded Kodi boxes in the Netherlands. The subsequent landmark ruling from the European Court of Justice will reverberate around Europe for years to come.

Behind the scenes, however, BREIN persistently tries to take much smaller operations offline, and not without success. Earlier this year it revealed it had taken down 231 illegal sites and services includes 84 linking sites, 63 streaming portals, and 34 torrent sites. Some of these shut down completely and others were forced to leave their hosting providers.

Much of this work flies under the radar but some current action, against an eBook site, is now being thrust into the public eye.

For more than five years, EBoek.info (eBook) has serviced Internet users looking to obtain comic books in Dutch. The site informs TorrentFreak it provides a legitimate service, targeted at people who have purchased a hard copy but also want their comics in digital format.

“EBoek.info is a site about comic books in the Dutch language. Besides some general information about the books, people who have legally obtained a hard copy of the books can find a link to an NZB file which enables them to download a digital version of the books they already have,” site representative ‘Zala’ says.

For those out of the loop, NZB files are a bit like Usenet’s version of .torrent files. They contain no copyrighted content themselves but do provide software clients with information on where to find specific content, so it can be downloaded to a user’s machine.

“BREIN claims that this is illegal as it is impossible for us to verify if our visitor is telling the truth [about having purchased a copy],” Zala reveals.

Speaking with TorrentFreak, BREIN chief Tim Kuik says there’s no question that offering downloads like this is illegal.

“It is plain and simple: the site makes links to unauthorized digital copies available to the general public and therefore is infringing copyright. It is distribution of the content without authorization of the rights holder,” Kuik says.

“The unauthorized copies are not private copies. The private copy exception does not apply to this kind of distribution. The private copy has not been made by the owner of the book himself for his own use. Someone else made the digital copy and is making it available to anyone who wants to download it provided he makes the unverified claim that he has a legal copy. This harms the normal exploitation of the

Zala says that BREIN has been trying to take his site offline for many years but more recently, the platform has utilized the services of Cloudflare, partly as a form of shield. As readers may be aware, a site behind Cloudflare has its originating IP addresses hidden from the public, not to mention BREIN, who values that kind of information. According to the operator, however, BREIN managed to obtain the information from the CDN provider.

“BREIN has tried for years to take our site offline. Recently, however, Cloudflare was so friendly to give them our IP address,” Zala notes.

A text copy of an email reportedly sent by BREIN to EBoek’s web host and seen by TF appears to confirm that Cloudflare handed over the information as suggested. Among other things, the email has BREIN informing the host that “The IP we got back from Cloudflare is XXX.XXX.XX.33.”

This means that BREIN was able to place direct pressure on EBoek.info’s web host, so only time will tell if that bears any fruit for the anti-piracy group. In the meantime, however, EBoek has decided to go public over its battle with BREIN.

“We have received a request from Stichting BREIN via our hosting provider to take EBoek.info offline,” the site informed its users yesterday.

Interestingly, it also appears that BREIN doesn’t appreciate that the operators of EBoek have failed to make their identities publicly known on their platform.

“The site operates anonymously which also is unlawful. Consumer protection requires that the owner/operator of a site identifies himself,” Kuik says.

According to EBoek, the anti-piracy outfit told the site’s web host that as a “commercial online service”, EBoek is required under EU law to display its “correct and complete business information” including names, addresses, and other information. But perhaps unsurprisingly, the site doesn’t want to play ball.

“In my opinion, you are confusing us with Facebook. They are a foreign commercial company with a European branch in Ireland, and therefore are subject to Irish legislation,” Zala says in an open letter to BREIN.

“Eboek.info, on the other hand, is a foreign hobby club with no commercial purpose, whose administrators have no connection with any country in the European Union. As administrators, we follow the laws of our country of residence which do not oblige us to disclose our identity through our website.

“The fact that Eboek is visible in the Netherlands does not just mean that we are going to adapt to Dutch rules, just as we don’t adapt the site to the rules of Saudi Arabia or China or wherever we are available.”

In a further snub to the anti-piracy group, EBoek says that all visitors to the site have to communicate with its operators via its guestbook, which is publicly visible.

“We see no reason to make an exception for Stichting BREIN,” the site notes.

What makes the situation more complex is that EBoek isn’t refusing dialog completely. The site says it doesn’t want to talk to BREIN but will speak to BREIN’s customers – the publishers of the comic books in question – noting that to date no complaints from publishers have ever been received.

While the parties argue about lines of communication, BREIN insists that following this year’s European Court of Justice decision in the GS Media case, a link to a known infringing work represents copyright infringement. In this case, an NZB file – which links to a location on Usenet – would generally fit the bill.

But despite focusing on the Dutch market, the operators of EBoek say the ruling doesn’t apply to them as they’re outside of the ECJ’s jurisdiction and aren’t commercially motivated. Refusing point blank to take their site offline, EBoek’s operators say that BREIN can do its worst, nothing will have much effect.

“[W]hat’s the worst thing that can happen? That our web host hands [BREIN] our address and IP data. In that case, it will turn out that…we are actually far away,” Zala says.

“[In the case the site goes offline], we’ll just put a backup on another server and, in this case, won’t make use of the ‘services’ of Cloudflare, the provider that apparently put BREIN on the right track.”

The question of jurisdiction is indeed an interesting one, particularly given BREIN’s focus in the Netherlands. But Kuik is clear – it is the area where the content is made available that matters.

“The law of the country where the content is made available applies. In this case the EU and amongst others the Netherlands,” Kuik concludes.

To be continued…..

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

Hollywood Giants Sue Kodi-powered ‘TickBox TV’ Over Piracy

Post Syndicated from Ernesto original https://torrentfreak.com/hollywood-giants-sue-kodi-powered-tickbox-tv-over-piracy-171014/

Online streaming piracy is booming and many people use dedicated media players to bring this content to their regular TVs.

The bare hardware is not illegal and neither is media player software such as Kodi. When these devices are loaded with copyright-infringing addons, however, they turn into an unprecedented piracy threat.

It becomes even more problematic when the sellers of these devices market their products as pirate tools. This is exactly what TickBox TV does, according to Hollywood’s major movie studios, Netflix, and Amazon.

TickBox is a Georgia-based provider of set-top boxes that allow users to stream a variety of popular media. The company’s devices use the Kodi media player and come with instructions on how to add various add-ons.

In a complaint filed in a California federal court yesterday, Universal, Columbia Pictures, Disney, 20th Century Fox, Paramount Pictures, Warner Bros, Amazon, and Netflix accuse Tickbox of inducing and contributing to copyright infringement.

“TickBox sells ‘TickBox TV,’ a computer hardware device that TickBox urges its customers to use as a tool for the mass infringement of Plaintiffs’ copyrighted motion pictures and television shows,” the complaint, picked up by THR, reads.

While the device itself does not host any infringing content, users are informed where they can find it.

The movie and TV studios stress that Tickbox’s marketing highlights its infringing uses with statements such as “if you’re tired of wasting money with online streaming services like Netflix, Hulu or Amazon Prime.”

Sick of paying high monthly fees?

“TickBox promotes the use of TickBox TV for overwhelmingly, if not exclusively, infringing purposes, and that is how its customers use TickBox TV. TickBox advertises TickBox TV as a substitute for authorized and legitimate distribution channels such as cable television or video-on-demand services like Amazon Prime and Netflix,” the studios’ lawyers write.

The complaint explains in detail how TickBox works. When users first boot up their device they are prompted to download the “TickBox TV Player” software. This comes with an instruction video guiding people to infringing streams.

“The TickBox TV instructional video urges the customer to use the ‘Select Your Theme’ button on the start-up menu for downloading addons. The ‘Themes’ are curated collections of popular addons that link to unauthorized streams of motion pictures and television shows.”

“Some of the most popular addons currently distributed — which are available through TickBox TV — are titled ‘Elysium,’ ‘Bob,’ and ‘Covenant’,” the complaint adds, showing screenshots of the interface.


The movie and TV studios, which are the founding members of the recently launched ACE anti-piracy initiative, want TickBox to stop selling their devices. In addition, they demand compensation for the damages they’ve suffered. Requesting the maximum statutory damages of $150,000 per copyright infringement, this can run into the millions.

The involvement of Amazon, albeit the content division, is notable since the online store itself sells dozens of similar streaming devices, some of which even list “infringing” addons.

The TickBox lawsuit is the first case in the United States where a group of major Hollywood players is targeting a streaming device. Earlier this year various Hollywood insiders voiced concerns about the piracy streaming epidemic and if this case goes their way, it probably won’t be the last.

A copy of the full complaint is available here (pdf)

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

Popular Zer0day Torrent Tracker Taken Offline By Mass Copyright Complaint

Post Syndicated from Andy original https://torrentfreak.com/popular-zer0day-torrent-tracker-taken-offline-by-mass-copyright-complaint-171014/

In January 2016, a BitTorrent enthusiast decided to launch a stand-alone tracker, purely for fun.

The Zer0day platform, which hosts no torrents, is a tracker in the purest sense, directing traffic between peers, no matter what content is involved and no matter where people are in the world.

With this type of tracker in short supply, it was soon utilized by The Pirate Bay and the now-defunct ExtraTorrent. By August 2016, it was tracking almost four million peers and a million torrents, a considerable contribution to the BitTorrent ecosystem.

After handling many ups and downs associated with a service of this type, the tracker eventually made it to the end of 2016 intact. This year it grew further still and by the end of September was tracking an impressive 5.5 million peers spread over 1.2 million torrents. Soon after, however, the tracker disappeared from the Internet without warning.

In an effort to find out what had happened, TorrentFreak contacted Zer0day’s operator who told us a familiar story. Without any warning at all, the site’s host pulled the plug on the service, despite having been paid 180 euros for hosting just a week earlier.

“We’re hereby informing you of the termination of your dedicated server due to a breach of our terms of service,” the host informed Zer0day.

“Hosting trackers on our servers that distribute infringing and copyrighted content is prohibited. This server was found to distribute such content. Should we identify additional similar activity in your services, we will be forced to close your account.”

While hosts tend not to worry too much about what their customers are doing, this one had just received a particularly lengthy complaint. Sent by the head of anti-piracy at French collecting society SCPP, it laid out the group’s problems with the Zer0day tracker.

“SCPP has been responsible for the collective management and protection of sound recordings and music videos producers’ rights since 1985. SCPP counts more than 2,600 members including the majority of independent French producers, in addition to independent European producers, and the major international companies: Sony, Universal and Warner,” the complaints reads.

“SCPP administers a catalog of 7,200,000 sound tracks and 77,000 music videos. SCPP is empowered by its members to take legal action in order to put an end to any infringements of the producers’ rights set out in Article L335-4 of the French Intellectual Property Code…..punishable by a three-year prison sentence or a fine of €300,000.”

Noting that it works on behalf of a number of labels and distributors including BMG, Sony Music, Universal Music, Warner Music and others, SCPP listed countless dozens of albums under its protection, each allegedly tracked by the Zer0day platform.

“It has come to our attention that these music albums are illegally being communicated to the public (made available for download) by various users of the BitTorrent-Network,” the complaint reads.

Noting that Zer0day is involved in the process, the anti-piracy outfit presented dozens of hash codes relating to protected works, demanding that the site stop facilitation of infringement on each and every one of them.

“We have proof that your tracker udp://tracker.zer0day.to:1337/announce provided peers of the BitTorrent-Network with information regarding these torrents, to be specific IP Addresses of peers that were offering without authorization the full albums for download, and that this information enabled peers to download files that contain the sound recordings to which our members producers have the exclusive rights.

“These sound recordings are thus being illegally communicated to the public, and your tracker is enabling the seeders to do so.”

Rather than take the hashes down from the tracker, SCPP actually demanded that Zer0day create a permanent blacklist within 24 hours, to ensure the corresponding torrents wouldn’t be tracked again.

“You should understand that this letter constitutes a notice to you that you may be liable for the infringing activity occurring on your service. In addition, if you ignore this notice, you may also be liable for any resulting infringement,” the complaint added.

But despite all the threats, SCPP didn’t receive the response they’d demanded since the operator of the site refused to take any action.

“Obviously, ‘info hashes’ are not copyrightable nor point to specific copyrighted content, or even have any meaning. Further, I cannot verify that request strings parameters (‘info hashes’) you sent me contain copyrighted material,” he told SCPP.

“Like the website says; for content removal kindly ask the indexing site to remove the listing and the .torrent file. Also, tracker software does not have an option to block request strings parameters (‘info hashes’).”

The net effect of non-compliance with SCPP was fairly dramatic and swift. Zer0day’s host took down the whole tracker instead and currently it remains offline. Whether it reappears depends on the site’s operator finding a suitable web host, but at the moment he says he has no idea where one will appear from.

“Currently I’m searching for some virtual private server as a temporary home for the tracker,” he concludes.

As mentioned in an earlier article detailing the problems sites like Zer0day.to face, trackers aren’t absolutely essential for the functioning of BitTorrent transfers. Nevertheless, their existence certainly improves matters for file-sharers so when they go down, millions can be affected.

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.

Backblaze Release 5.1 – RMM Compatibility for Mass Deployments

Post Syndicated from Yev original https://www.backblaze.com/blog/rmm-for-mass-deployments/

diagram of Backblaze remote monitoring and management

Introducing Backblaze Computer Backup Release 5.1

This is a relatively minor release in terms of the core Backblaze Computer Backup service functionality, but is a big deal for Backblaze for Business as we’ve updated our Mac and PC clients to be RMM (Remote Monitoring and Management) compatible.

What Is New?

  • Updated Mac and PC clients to better handle large file uploads
  • Updated PC downloader to improve stability
  • Added RMM support for PC and Mac clients

What Is RMM?

RMM stands for “Remote Monitoring and Management.” It’s a way to administer computers that might be distributed geographically, without having access to the actual machine. If you are a systems administrator working with anywhere from a few distributed computers to a few thousand, you’re familiar with RMM and how it makes life easier.

The new clients allow administrators to deploy Backblaze Computer Backup through most “silent” installation/mass deployment tools. Two popular RMM tools are Munki and Jamf. We’ve written up knowledge base articles for both of these.

munki logo jamf logo
Learn more about Munki Learn more about Jamf

Do I Need To Use RMM Tools?

No — unless you are a systems administrator or someone who is deploying Backblaze to a lot of people all at once, you do not have to worry about RMM support.

Release Version Number:

Mac:  5.1.0
PC:  5.1.0


October 12, 2017

Upgrade Methods:

  • “Check for Updates” on the Backblaze Client (right click on the Backblaze icon and then select “Check for Updates”)
  • Download from: https://secure.backblaze.com/update.htm
  • Auto-update will begin in a couple of weeks
Mac backup update PC backup update
Updating Backblaze on Mac Updating Backblaze on Windows


If you have any questions, please contact Backblaze Support at www.backblaze.com/help.

The post Backblaze Release 5.1 – RMM Compatibility for Mass Deployments appeared first on Backblaze Blog | Cloud Storage & Cloud Backup.

Predict Billboard Top 10 Hits Using RStudio, H2O and Amazon Athena

Post Syndicated from Gopal Wunnava original https://aws.amazon.com/blogs/big-data/predict-billboard-top-10-hits-using-rstudio-h2o-and-amazon-athena/

Success in the popular music industry is typically measured in terms of the number of Top 10 hits artists have to their credit. The music industry is a highly competitive multi-billion dollar business, and record labels incur various costs in exchange for a percentage of the profits from sales and concert tickets.

Predicting the success of an artist’s release in the popular music industry can be difficult. One release may be extremely popular, resulting in widespread play on TV, radio and social media, while another single may turn out quite unpopular, and therefore unprofitable. Record labels need to be selective in their decision making, and predictive analytics can help them with decision making around the type of songs and artists they need to promote.

In this walkthrough, you leverage H2O.ai, Amazon Athena, and RStudio to make predictions on whether a song might make it to the Top 10 Billboard charts. You explore the GLM, GBM, and deep learning modeling techniques using H2O’s rapid, distributed and easy-to-use open source parallel processing engine. RStudio is a popular IDE, licensed either commercially or under AGPLv3, for working with R. This is ideal if you don’t want to connect to a server via SSH and use code editors such as vi to do analytics. RStudio is available in a desktop version, or a server version that allows you to access R via a web browser. RStudio’s Notebooks feature is used to demonstrate the execution of code and output. In addition, this post showcases how you can leverage Athena for query and interactive analysis during the modeling phase. A working knowledge of statistics and machine learning would be helpful to interpret the analysis being performed in this post.


Your goal is to predict whether a song will make it to the Top 10 Billboard charts. For this purpose, you will be using multiple modeling techniques―namely GLM, GBM and deep learning―and choose the model that is the best fit.

This solution involves the following steps:

  • Install and configure RStudio with Athena
  • Log in to RStudio
  • Install R packages
  • Connect to Athena
  • Create a dataset
  • Create models

Install and configure RStudio with Athena

Use the following AWS CloudFormation stack to install, configure, and connect RStudio on an Amazon EC2 instance with Athena.

Launching this stack creates all required resources and prerequisites:

  • Amazon EC2 instance with Amazon Linux (minimum size of t2.large is recommended)
  • Provisioning of the EC2 instance in an existing VPC and public subnet
  • Installation of Java 8
  • Assignment of an IAM role to the EC2 instance with the required permissions for accessing Athena and Amazon S3
  • Security group allowing access to the RStudio and SSH ports from the internet (I recommend restricting access to these ports)
  • S3 staging bucket required for Athena (referenced within RStudio as ATHENABUCKET)
  • RStudio username and password
  • Setup logs in Amazon CloudWatch Logs (if needed for additional troubleshooting)
  • Amazon EC2 Systems Manager agent, which makes it easy to manage and patch

All AWS resources are created in the US-East-1 Region. To avoid cross-region data transfer fees, launch the CloudFormation stack in the same region. To check the availability of Athena in other regions, see Region Table.

Log in to RStudio

The instance security group has been automatically configured to allow incoming connections on the RStudio port 8787 from any source internet address. You can edit the security group to restrict source IP access. If you have trouble connecting, ensure that port 8787 isn’t blocked by subnet network ACLS or by your outgoing proxy/firewall.

  1. In the CloudFormation stack, choose Outputs, Value, and then open the RStudio URL. You might need to wait for a few minutes until the instance has been launched.
  2. Log in to RStudio with the and password you provided during setup.

Install R packages

Next, install the required R packages from the RStudio console. You can download the R notebook file containing just the code.

#install pacman – a handy package manager for managing installs
if("pacman" %in% rownames(installed.packages()) == FALSE)
h2o.init(nthreads = -1)
##  Connection successful!
## R is connected to the H2O cluster: 
##     H2O cluster uptime:         2 hours 42 minutes 
##     H2O cluster version: 
##     H2O cluster version age:    4 months and 4 days !!! 
##     H2O cluster name:           H2O_started_from_R_rstudio_hjx881 
##     H2O cluster total nodes:    1 
##     H2O cluster total memory:   3.30 GB 
##     H2O cluster total cores:    4 
##     H2O cluster allowed cores:  4 
##     H2O cluster healthy:        TRUE 
##     H2O Connection ip:          localhost 
##     H2O Connection port:        54321 
##     H2O Connection proxy:       NA 
##     H2O Internal Security:      FALSE 
##     R Version:                  R version 3.3.3 (2017-03-06)
## Warning in h2o.clusterInfo(): 
## Your H2O cluster version is too old (4 months and 4 days)!
## Please download and install the latest version from http://h2o.ai/download/
#install aws sdk if not present (pre-requisite for using Athena with an IAM role)
if (!aws_sdk_present()) {


Connect to Athena

Next, establish a connection to Athena from RStudio, using an IAM role associated with your EC2 instance. Use ATHENABUCKET to specify the S3 staging directory.

URL <- 'https://s3.amazonaws.com/athena-downloads/drivers/AthenaJDBC41-1.0.1.jar'
fil <- basename(URL)
#download the file into current working directory
if (!file.exists(fil)) download.file(URL, fil)
#verify that the file has been downloaded successfully
## [1] "AthenaJDBC41-1.0.1.jar"
drv <- JDBC(driverClass="com.amazonaws.athena.jdbc.AthenaDriver", fil, identifier.quote="'")

con <- jdbcConnection <- dbConnect(drv, 'jdbc:awsathena://athena.us-east-1.amazonaws.com:443/',

Verify the connection. The results returned depend on your specific Athena setup.

## <JDBCConnection>
##  [1] "gdelt"               "wikistats"           "elb_logs_raw_native"
##  [4] "twitter"             "twitter2"            "usermovieratings"   
##  [7] "eventcodes"          "events"              "billboard"          
## [10] "billboardtop10"      "elb_logs"            "gdelthist"          
## [13] "gdeltmaster"         "twitter"             "twitter3"

Create a dataset

For this analysis, you use a sample dataset combining information from Billboard and Wikipedia with Echo Nest data in the Million Songs Dataset. Upload this dataset into your own S3 bucket. The table below provides a description of the fields used in this dataset.

Field Description
year Year that song was released
songtitle Title of the song
artistname Name of the song artist
songid Unique identifier for the song
artistid Unique identifier for the song artist
timesignature Variable estimating the time signature of the song
timesignature_confidence Confidence in the estimate for the timesignature
loudness Continuous variable indicating the average amplitude of the audio in decibels
tempo Variable indicating the estimated beats per minute of the song
tempo_confidence Confidence in the estimate for tempo
key Variable with twelve levels indicating the estimated key of the song (C, C#, B)
key_confidence Confidence in the estimate for key
energy Variable that represents the overall acoustic energy of the song, using a mix of features such as loudness
pitch Continuous variable that indicates the pitch of the song
timbre_0_min thru timbre_11_min Variables that indicate the minimum values over all segments for each of the twelve values in the timbre vector
timbre_0_max thru timbre_11_max Variables that indicate the maximum values over all segments for each of the twelve values in the timbre vector
top10 Indicator for whether or not the song made it to the Top 10 of the Billboard charts (1 if it was in the top 10, and 0 if not)

Create an Athena table based on the dataset

In the Athena console, select the default database, sampled, or create a new database.

Run the following create table statement.

create external table if not exists billboard
year int,
songtitle string,
artistname string,
songID string,
artistID string,
timesignature int,
timesignature_confidence double,
loudness double,
tempo double,
tempo_confidence double,
key int,
key_confidence double,
energy double,
pitch double,
timbre_0_min double,
timbre_0_max double,
timbre_1_min double,
timbre_1_max double,
timbre_2_min double,
timbre_2_max double,
timbre_3_min double,
timbre_3_max double,
timbre_4_min double,
timbre_4_max double,
timbre_5_min double,
timbre_5_max double,
timbre_6_min double,
timbre_6_max double,
timbre_7_min double,
timbre_7_max double,
timbre_8_min double,
timbre_8_max double,
timbre_9_min double,
timbre_9_max double,
timbre_10_min double,
timbre_10_max double,
timbre_11_min double,
timbre_11_max double,
Top10 int
LOCATION 's3://aws-bigdata-blog/artifacts/predict-billboard/data'

Inspect the table definition for the ‘billboard’ table that you have created. If you chose a database other than sampledb, replace that value with your choice.

dbGetQuery(con, "show create table sampledb.billboard")
##                                      createtab_stmt
## 1       CREATE EXTERNAL TABLE `sampledb.billboard`(
## 2                                       `year` int,
## 3                               `songtitle` string,
## 4                              `artistname` string,
## 5                                  `songid` string,
## 6                                `artistid` string,
## 7                              `timesignature` int,
## 8                `timesignature_confidence` double,
## 9                                `loudness` double,
## 10                                  `tempo` double,
## 11                       `tempo_confidence` double,
## 12                                       `key` int,
## 13                         `key_confidence` double,
## 14                                 `energy` double,
## 15                                  `pitch` double,
## 16                           `timbre_0_min` double,
## 17                           `timbre_0_max` double,
## 18                           `timbre_1_min` double,
## 19                           `timbre_1_max` double,
## 20                           `timbre_2_min` double,
## 21                           `timbre_2_max` double,
## 22                           `timbre_3_min` double,
## 23                           `timbre_3_max` double,
## 24                           `timbre_4_min` double,
## 25                           `timbre_4_max` double,
## 26                           `timbre_5_min` double,
## 27                           `timbre_5_max` double,
## 28                           `timbre_6_min` double,
## 29                           `timbre_6_max` double,
## 30                           `timbre_7_min` double,
## 31                           `timbre_7_max` double,
## 32                           `timbre_8_min` double,
## 33                           `timbre_8_max` double,
## 34                           `timbre_9_min` double,
## 35                           `timbre_9_max` double,
## 36                          `timbre_10_min` double,
## 37                          `timbre_10_max` double,
## 38                          `timbre_11_min` double,
## 39                          `timbre_11_max` double,
## 40                                     `top10` int)
## 41                             ROW FORMAT DELIMITED 
## 42                         FIELDS TERMINATED BY ',' 
## 43                            STORED AS INPUTFORMAT 
## 44       'org.apache.hadoop.mapred.TextInputFormat' 
## 45                                     OUTPUTFORMAT 
## 46  'org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat'
## 47                                        LOCATION
## 48    's3://aws-bigdata-blog/artifacts/predict-billboard/data'
## 49                                  TBLPROPERTIES (
## 50            'transient_lastDdlTime'='1505484133')

Run a sample query

Next, run a sample query to obtain a list of all songs from Janet Jackson that made it to the Billboard Top 10 charts.

dbGetQuery(con, " SELECT songtitle,artistname,top10   FROM sampledb.billboard WHERE lower(artistname) =     'janet jackson' AND top10 = 1")
##                       songtitle    artistname top10
## 1                       Runaway Janet Jackson     1
## 2               Because Of Love Janet Jackson     1
## 3                         Again Janet Jackson     1
## 4                            If Janet Jackson     1
## 5  Love Will Never Do (Without You) Janet Jackson 1
## 6                     Black Cat Janet Jackson     1
## 7               Come Back To Me Janet Jackson     1
## 8                       Alright Janet Jackson     1
## 9                      Escapade Janet Jackson     1
## 10                Rhythm Nation Janet Jackson     1

Determine how many songs in this dataset are specifically from the year 2010.

dbGetQuery(con, " SELECT count(*)   FROM sampledb.billboard WHERE year = 2010")
##   _col0
## 1   373

The sample dataset provides certain song properties of interest that can be analyzed to gauge the impact to the song’s overall popularity. Look at one such property, timesignature, and determine the value that is the most frequent among songs in the database. Timesignature is a measure of the number of beats and the type of note involved.

Running the query directly may result in an error, as shown in the commented lines below. This error is a result of trying to retrieve a large result set over a JDBC connection, which can cause out-of-memory issues at the client level. To address this, reduce the fetch size and run again.

#t<-dbGetQuery(con, " SELECT timesignature FROM sampledb.billboard")
#Note:  Running the preceding query results in the following error: 
#Error in .jcall(rp, "I", "fetch", stride, block): java.sql.SQLException: The requested #fetchSize is more than the allowed value in Athena. Please reduce the fetchSize and try #again. Refer to the Athena documentation for valid fetchSize values.
# Use the dbSendQuery function, reduce the fetch size, and run again
r <- dbSendQuery(con, " SELECT timesignature     FROM sampledb.billboard")
dftimesignature<- fetch(r, n=-1, block=100)
## [1] TRUE
## dftimesignature
##    0    1    3    4    5    7 
##   10  143  503 6787  112   19
## [1] 7574

From the results, observe that 6787 songs have a timesignature of 4.

Next, determine the song with the highest tempo.

dbGetQuery(con, " SELECT songtitle,artistname,tempo   FROM sampledb.billboard WHERE tempo = (SELECT max(tempo) FROM sampledb.billboard) ")
##                   songtitle      artistname   tempo
## 1 Wanna Be Startin' Somethin' Michael Jackson 244.307

Create the training dataset

Your model needs to be trained such that it can learn and make accurate predictions. Split the data into training and test datasets, and create the training dataset first.  This dataset contains all observations from the year 2009 and earlier. You may face the same JDBC connection issue pointed out earlier, so this query uses a fetch size.

#BillboardTrain <- dbGetQuery(con, "SELECT * FROM sampledb.billboard WHERE year <= 2009")
#Running the preceding query results in the following error:-
#Error in .verify.JDBC.result(r, "Unable to retrieve JDBC result set for ", : Unable to retrieve #JDBC result set for SELECT * FROM sampledb.billboard WHERE year <= 2009 (Internal error)
#Follow the same approach as before to address this issue.

r <- dbSendQuery(con, "SELECT * FROM sampledb.billboard WHERE year <= 2009")
BillboardTrain <- fetch(r, n=-1, block=100)
## [1] TRUE
##   year           songtitle artistname timesignature
## 1 2009 The Awkward Goodbye    Athlete             3
## 2 2009        Rubik's Cube    Athlete             3
##   timesignature_confidence loudness   tempo tempo_confidence
## 1                    0.732   -6.320  89.614   0.652
## 2                    0.906   -9.541 117.742   0.542
## [1] 7201

Create the test dataset

BillboardTest <- dbGetQuery(con, "SELECT * FROM sampledb.billboard where year = 2010")
##   year              songtitle        artistname key
## 1 2010 This Is the House That Doubt Built A Day to Remember  11
## 2 2010        Sticks & Bricks A Day to Remember  10
##   key_confidence    energy pitch timbre_0_min
## 1          0.453 0.9666556 0.024        0.002
## 2          0.469 0.9847095 0.025        0.000
## [1] 373

Convert the training and test datasets into H2O dataframes

train.h2o <- as.h2o(BillboardTrain)
  |                                                                 |   0%
  |=================================================================| 100%
test.h2o <- as.h2o(BillboardTest)
  |                                                                 |   0%
  |=================================================================| 100%

Inspect the column names in your H2O dataframes.

##  [1] "year"                     "songtitle"               
##  [3] "artistname"               "songid"                  
##  [5] "artistid"                 "timesignature"           
##  [7] "timesignature_confidence" "loudness"                
##  [9] "tempo"                    "tempo_confidence"        
## [11] "key"                      "key_confidence"          
## [13] "energy"                   "pitch"                   
## [15] "timbre_0_min"             "timbre_0_max"            
## [17] "timbre_1_min"             "timbre_1_max"            
## [19] "timbre_2_min"             "timbre_2_max"            
## [21] "timbre_3_min"             "timbre_3_max"            
## [23] "timbre_4_min"             "timbre_4_max"            
## [25] "timbre_5_min"             "timbre_5_max"            
## [27] "timbre_6_min"             "timbre_6_max"            
## [29] "timbre_7_min"             "timbre_7_max"            
## [31] "timbre_8_min"             "timbre_8_max"            
## [33] "timbre_9_min"             "timbre_9_max"            
## [35] "timbre_10_min"            "timbre_10_max"           
## [37] "timbre_11_min"            "timbre_11_max"           
## [39] "top10"

Create models

You need to designate the independent and dependent variables prior to applying your modeling algorithms. Because you’re trying to predict the ‘top10’ field, this would be your dependent variable and everything else would be independent.

Create your first model using GLM. Because GLM works best with numeric data, you create your model by dropping non-numeric variables. You only use the variables in the dataset that describe the numerical attributes of the song in the logistic regression model. You won’t use these variables:  “year”, “songtitle”, “artistname”, “songid”, or “artistid”.

y.dep <- 39
x.indep <- c(6:38)
##  [1]  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
## [24] 29 30 31 32 33 34 35 36 37 38

Create Model 1: All numeric variables

Create Model 1 with the training dataset, using GLM as the modeling algorithm and H2O’s built-in h2o.glm function.

modelh1 <- h2o.glm( y = y.dep, x = x.indep, training_frame = train.h2o, family = "binomial")
  |                                                                 |   0%
  |=====                                                            |   8%
  |=================================================================| 100%

Measure the performance of Model 1, using H2O’s built-in performance function.

## H2OBinomialMetrics: glm
## MSE:  0.09924684
## RMSE:  0.3150347
## LogLoss:  0.3220267
## Mean Per-Class Error:  0.2380168
## AUC:  0.8431394
## Gini:  0.6862787
## R^2:  0.254663
## Null Deviance:  326.0801
## Residual Deviance:  240.2319
## AIC:  308.2319
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
##          0   1    Error     Rate
## 0      255  59 0.187898  =59/314
## 1       17  42 0.288136   =17/59
## Totals 272 101 0.203753  =76/373
## Maximum Metrics: Maximum metrics at their respective thresholds
##                         metric threshold    value idx
## 1                       max f1  0.192772 0.525000 100
## 2                       max f2  0.124912 0.650510 155
## 3                 max f0point5  0.416258 0.612903  23
## 4                 max accuracy  0.416258 0.879357  23
## 5                max precision  0.813396 1.000000   0
## 6                   max recall  0.037579 1.000000 282
## 7              max specificity  0.813396 1.000000   0
## 8             max absolute_mcc  0.416258 0.455251  23
## 9   max min_per_class_accuracy  0.161402 0.738854 125
## 10 max mean_per_class_accuracy  0.124912 0.765006 155
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or ` 
## [1] 0.8431394

The AUC metric provides insight into how well the classifier is able to separate the two classes. In this case, the value of 0.8431394 indicates that the classification is good. (A value of 0.5 indicates a worthless test, while a value of 1.0 indicates a perfect test.)

Next, inspect the coefficients of the variables in the dataset.

dfmodelh1 <- as.data.frame(h2o.varimp(modelh1))
##                       names coefficients sign
## 1              timbre_0_max  1.290938663  NEG
## 2                  loudness  1.262941934  POS
## 3                     pitch  0.616995941  NEG
## 4              timbre_1_min  0.422323735  POS
## 5              timbre_6_min  0.349016024  NEG
## 6                    energy  0.348092062  NEG
## 7             timbre_11_min  0.307331997  NEG
## 8              timbre_3_max  0.302225619  NEG
## 9             timbre_11_max  0.243632060  POS
## 10             timbre_4_min  0.224233951  POS
## 11             timbre_4_max  0.204134342  POS
## 12             timbre_5_min  0.199149324  NEG
## 13             timbre_0_min  0.195147119  POS
## 14 timesignature_confidence  0.179973904  POS
## 15         tempo_confidence  0.144242598  POS
## 16            timbre_10_max  0.137644568  POS
## 17             timbre_7_min  0.126995955  NEG
## 18            timbre_10_min  0.123851179  POS
## 19             timbre_7_max  0.100031481  NEG
## 20             timbre_2_min  0.096127636  NEG
## 21           key_confidence  0.083115820  POS
## 22             timbre_6_max  0.073712419  POS
## 23            timesignature  0.067241917  POS
## 24             timbre_8_min  0.061301881  POS
## 25             timbre_8_max  0.060041698  POS
## 26                      key  0.056158445  POS
## 27             timbre_3_min  0.050825116  POS
## 28             timbre_9_max  0.033733561  POS
## 29             timbre_2_max  0.030939072  POS
## 30             timbre_9_min  0.020708113  POS
## 31             timbre_1_max  0.014228818  NEG
## 32                    tempo  0.008199861  POS
## 33             timbre_5_max  0.004837870  POS
## 34                                    NA <NA>

Typically, songs with heavier instrumentation tend to be louder (have higher values in the variable “loudness”) and more energetic (have higher values in the variable “energy”). This knowledge is helpful for interpreting the modeling results.

You can make the following observations from the results:

  • The coefficient estimates for the confidence values associated with the time signature, key, and tempo variables are positive. This suggests that higher confidence leads to a higher predicted probability of a Top 10 hit.
  • The coefficient estimate for loudness is positive, meaning that mainstream listeners prefer louder songs with heavier instrumentation.
  • The coefficient estimate for energy is negative, meaning that mainstream listeners prefer songs that are less energetic, which are those songs with light instrumentation.

These coefficients lead to contradictory conclusions for Model 1. This could be due to multicollinearity issues. Inspect the correlation between the variables “loudness” and “energy” in the training set.

## [1] 0.7399067

This number indicates that these two variables are highly correlated, and Model 1 does indeed suffer from multicollinearity. Typically, you associate a value of -1.0 to -0.5 or 1.0 to 0.5 to indicate strong correlation, and a value of 0.1 to 0.1 to indicate weak correlation. To avoid this correlation issue, omit one of these two variables and re-create the models.

You build two variations of the original model:

  • Model 2, in which you keep “energy” and omit “loudness”
  • Model 3, in which you keep “loudness” and omit “energy”

You compare these two models and choose the model with a better fit for this use case.

Create Model 2: Keep energy and omit loudness

##  [1] "year"                     "songtitle"               
##  [3] "artistname"               "songid"                  
##  [5] "artistid"                 "timesignature"           
##  [7] "timesignature_confidence" "loudness"                
##  [9] "tempo"                    "tempo_confidence"        
## [11] "key"                      "key_confidence"          
## [13] "energy"                   "pitch"                   
## [15] "timbre_0_min"             "timbre_0_max"            
## [17] "timbre_1_min"             "timbre_1_max"            
## [19] "timbre_2_min"             "timbre_2_max"            
## [21] "timbre_3_min"             "timbre_3_max"            
## [23] "timbre_4_min"             "timbre_4_max"            
## [25] "timbre_5_min"             "timbre_5_max"            
## [27] "timbre_6_min"             "timbre_6_max"            
## [29] "timbre_7_min"             "timbre_7_max"            
## [31] "timbre_8_min"             "timbre_8_max"            
## [33] "timbre_9_min"             "timbre_9_max"            
## [35] "timbre_10_min"            "timbre_10_max"           
## [37] "timbre_11_min"            "timbre_11_max"           
## [39] "top10"
y.dep <- 39
x.indep <- c(6:7,9:38)
##  [1]  6  7  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
## [24] 30 31 32 33 34 35 36 37 38
modelh2 <- h2o.glm( y = y.dep, x = x.indep, training_frame = train.h2o, family = "binomial")
  |                                                                 |   0%
  |=======                                                          |  10%
  |=================================================================| 100%

Measure the performance of Model 2.

## H2OBinomialMetrics: glm
## MSE:  0.09922606
## RMSE:  0.3150017
## LogLoss:  0.3228213
## Mean Per-Class Error:  0.2490554
## AUC:  0.8431933
## Gini:  0.6863867
## R^2:  0.2548191
## Null Deviance:  326.0801
## Residual Deviance:  240.8247
## AIC:  306.8247
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
##          0  1    Error     Rate
## 0      280 34 0.108280  =34/314
## 1       23 36 0.389831   =23/59
## Totals 303 70 0.152815  =57/373
## Maximum Metrics: Maximum metrics at their respective thresholds
##                         metric threshold    value idx
## 1                       max f1  0.254391 0.558140  69
## 2                       max f2  0.113031 0.647208 157
## 3                 max f0point5  0.413999 0.596026  22
## 4                 max accuracy  0.446250 0.876676  18
## 5                max precision  0.811739 1.000000   0
## 6                   max recall  0.037682 1.000000 283
## 7              max specificity  0.811739 1.000000   0
## 8             max absolute_mcc  0.254391 0.469060  69
## 9   max min_per_class_accuracy  0.141051 0.716561 131
## 10 max mean_per_class_accuracy  0.113031 0.761821 157
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or `h2o.gainsLift(<model>, valid=<T/F>, xval=<T/F>)`
dfmodelh2 <- as.data.frame(h2o.varimp(modelh2))
##                       names coefficients sign
## 1                     pitch  0.700331511  NEG
## 2              timbre_1_min  0.510270513  POS
## 3              timbre_0_max  0.402059546  NEG
## 4              timbre_6_min  0.333316236  NEG
## 5             timbre_11_min  0.331647383  NEG
## 6              timbre_3_max  0.252425901  NEG
## 7             timbre_11_max  0.227500308  POS
## 8              timbre_4_max  0.210663865  POS
## 9              timbre_0_min  0.208516163  POS
## 10             timbre_5_min  0.202748055  NEG
## 11             timbre_4_min  0.197246582  POS
## 12            timbre_10_max  0.172729619  POS
## 13         tempo_confidence  0.167523934  POS
## 14 timesignature_confidence  0.167398830  POS
## 15             timbre_7_min  0.142450727  NEG
## 16             timbre_8_max  0.093377516  POS
## 17            timbre_10_min  0.090333426  POS
## 18            timesignature  0.085851625  POS
## 19             timbre_7_max  0.083948442  NEG
## 20           key_confidence  0.079657073  POS
## 21             timbre_6_max  0.076426046  POS
## 22             timbre_2_min  0.071957831  NEG
## 23             timbre_9_max  0.071393189  POS
## 24             timbre_8_min  0.070225578  POS
## 25                      key  0.061394702  POS
## 26             timbre_3_min  0.048384697  POS
## 27             timbre_1_max  0.044721121  NEG
## 28                   energy  0.039698433  POS
## 29             timbre_5_max  0.039469064  POS
## 30             timbre_2_max  0.018461133  POS
## 31                    tempo  0.013279926  POS
## 32             timbre_9_min  0.005282143  NEG
## 33                                    NA <NA>

## [1] 0.8431933

You can make the following observations:

  • The AUC metric is 0.8431933.
  • Inspecting the coefficient of the variable energy, Model 2 suggests that songs with high energy levels tend to be more popular. This is as per expectation.
  • As H2O orders variables by significance, the variable energy is not significant in this model.

You can conclude that Model 2 is not ideal for this use , as energy is not significant.

CreateModel 3: Keep loudness but omit energy

##  [1] "year"                     "songtitle"               
##  [3] "artistname"               "songid"                  
##  [5] "artistid"                 "timesignature"           
##  [7] "timesignature_confidence" "loudness"                
##  [9] "tempo"                    "tempo_confidence"        
## [11] "key"                      "key_confidence"          
## [13] "energy"                   "pitch"                   
## [15] "timbre_0_min"             "timbre_0_max"            
## [17] "timbre_1_min"             "timbre_1_max"            
## [19] "timbre_2_min"             "timbre_2_max"            
## [21] "timbre_3_min"             "timbre_3_max"            
## [23] "timbre_4_min"             "timbre_4_max"            
## [25] "timbre_5_min"             "timbre_5_max"            
## [27] "timbre_6_min"             "timbre_6_max"            
## [29] "timbre_7_min"             "timbre_7_max"            
## [31] "timbre_8_min"             "timbre_8_max"            
## [33] "timbre_9_min"             "timbre_9_max"            
## [35] "timbre_10_min"            "timbre_10_max"           
## [37] "timbre_11_min"            "timbre_11_max"           
## [39] "top10"
y.dep <- 39
x.indep <- c(6:12,14:38)
##  [1]  6  7  8  9 10 11 12 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
## [24] 30 31 32 33 34 35 36 37 38
modelh3 <- h2o.glm( y = y.dep, x = x.indep, training_frame = train.h2o, family = "binomial")
  |                                                                 |   0%
  |========                                                         |  12%
  |=================================================================| 100%
## H2OBinomialMetrics: glm
## MSE:  0.0978859
## RMSE:  0.3128672
## LogLoss:  0.3178367
## Mean Per-Class Error:  0.264925
## AUC:  0.8492389
## Gini:  0.6984778
## R^2:  0.2648836
## Null Deviance:  326.0801
## Residual Deviance:  237.1062
## AIC:  303.1062
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
##          0  1    Error     Rate
## 0      286 28 0.089172  =28/314
## 1       26 33 0.440678   =26/59
## Totals 312 61 0.144772  =54/373
## Maximum Metrics: Maximum metrics at their respective thresholds
##                         metric threshold    value idx
## 1                       max f1  0.273799 0.550000  60
## 2                       max f2  0.125503 0.663265 155
## 3                 max f0point5  0.435479 0.628931  24
## 4                 max accuracy  0.435479 0.882038  24
## 5                max precision  0.821606 1.000000   0
## 6                   max recall  0.038328 1.000000 280
## 7              max specificity  0.821606 1.000000   0
## 8             max absolute_mcc  0.435479 0.471426  24
## 9   max min_per_class_accuracy  0.173693 0.745763 120
## 10 max mean_per_class_accuracy  0.125503 0.775073 155
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or `h2o.gainsLift(<model>, valid=<T/F>, xval=<T/F>)`
dfmodelh3 <- as.data.frame(h2o.varimp(modelh3))
##                       names coefficients sign
## 1              timbre_0_max 1.216621e+00  NEG
## 2                  loudness 9.780973e-01  POS
## 3                     pitch 7.249788e-01  NEG
## 4              timbre_1_min 3.891197e-01  POS
## 5              timbre_6_min 3.689193e-01  NEG
## 6             timbre_11_min 3.086673e-01  NEG
## 7              timbre_3_max 3.025593e-01  NEG
## 8             timbre_11_max 2.459081e-01  POS
## 9              timbre_4_min 2.379749e-01  POS
## 10             timbre_4_max 2.157627e-01  POS
## 11             timbre_0_min 1.859531e-01  POS
## 12             timbre_5_min 1.846128e-01  NEG
## 13 timesignature_confidence 1.729658e-01  POS
## 14             timbre_7_min 1.431871e-01  NEG
## 15            timbre_10_max 1.366703e-01  POS
## 16            timbre_10_min 1.215954e-01  POS
## 17         tempo_confidence 1.183698e-01  POS
## 18             timbre_2_min 1.019149e-01  NEG
## 19           key_confidence 9.109701e-02  POS
## 20             timbre_7_max 8.987908e-02  NEG
## 21             timbre_6_max 6.935132e-02  POS
## 22             timbre_8_max 6.878241e-02  POS
## 23            timesignature 6.120105e-02  POS
## 24                      key 5.814805e-02  POS
## 25             timbre_8_min 5.759228e-02  POS
## 26             timbre_1_max 2.930285e-02  NEG
## 27             timbre_9_max 2.843755e-02  POS
## 28             timbre_3_min 2.380245e-02  POS
## 29             timbre_2_max 1.917035e-02  POS
## 30             timbre_5_max 1.715813e-02  POS
## 31                    tempo 1.364418e-02  NEG
## 32             timbre_9_min 8.463143e-05  NEG
## 33                                    NA <NA>
## Warning in h2o.find_row_by_threshold(object, t): Could not find exact
## threshold: 0.5 for this set of metrics; using closest threshold found:
## 0.501855569251422. Run `h2o.predict` and apply your desired threshold on a
## probability column.
## [[1]]
## [1] 0.2033898
## [1] 0.8492389

You can make the following observations:

  • The AUC metric is 0.8492389.
  • From the confusion matrix, the model correctly predicts that 33 songs will be top 10 hits (true positives). However, it has 26 false positives (songs that the model predicted would be Top 10 hits, but ended up not being Top 10 hits).
  • Loudness has a positive coefficient estimate, meaning that this model predicts that songs with heavier instrumentation tend to be more popular. This is the same conclusion from Model 2.
  • Loudness is significant in this model.

Overall, Model 3 predicts a higher number of top 10 hits with an accuracy rate that is acceptable. To choose the best fit for production runs, record labels should consider the following factors:

  • Desired model accuracy at a given threshold
  • Number of correct predictions for top10 hits
  • Tolerable number of false positives or false negatives

Next, make predictions using Model 3 on the test dataset.

predict.regh <- h2o.predict(modelh3, test.h2o)
  |                                                                 |   0%
  |=================================================================| 100%
##   predict        p0          p1
## 1       0 0.9654739 0.034526052
## 2       0 0.9654748 0.034525236
## 3       0 0.9635547 0.036445318
## 4       0 0.9343579 0.065642149
## 5       0 0.9978334 0.002166601
## 6       0 0.9779949 0.022005078
## [373 rows x 3 columns]
##   predict
## 1       0
## 2       0
## 3       0
## 4       0
## 5       0
## 6       0
## [373 rows x 1 column]
#Rename the predicted column 
colnames(dpr)[colnames(dpr) == 'predict'] <- 'predict_top10'
##   0   1 
## 312  61

The first set of output results specifies the probabilities associated with each predicted observation.  For example, observation 1 is 96.54739% likely to not be a Top 10 hit, and 3.4526052% likely to be a Top 10 hit (predict=1 indicates Top 10 hit and predict=0 indicates not a Top 10 hit).  The second set of results list the actual predictions made.  From the third set of results, this model predicts that 61 songs will be top 10 hits.

Compute the baseline accuracy, by assuming that the baseline predicts the most frequent outcome, which is that most songs are not Top 10 hits.

##   0   1 
## 314  59

Now observe that the baseline model would get 314 observations correct, and 59 wrong, for an accuracy of 314/(314+59) = 0.8418231.

It seems that Model 3, with an accuracy of 0.8552, provides you with a small improvement over the baseline model. But is this model useful for record labels?

View the two models from an investment perspective:

  • A production company is interested in investing in songs that are more likely to make it to the Top 10. The company’s objective is to minimize the risk of financial losses attributed to investing in songs that end up unpopular.
  • How many songs does Model 3 correctly predict as a Top 10 hit in 2010? Looking at the confusion matrix, you see that it predicts 33 top 10 hits correctly at an optimal threshold, which is more than half the number
  • It will be more useful to the record label if you can provide the production company with a list of songs that are highly likely to end up in the Top 10.
  • The baseline model is not useful, as it simply does not label any song as a hit.

Considering the three models built so far, you can conclude that Model 3 proves to be the best investment choice for the record label.

GBM model

H2O provides you with the ability to explore other learning models, such as GBM and deep learning. Explore building a model using the GBM technique, using the built-in h2o.gbm function.

Before you do this, you need to convert the target variable to a factor for multinomial classification techniques.

gbm.modelh <- h2o.gbm(y=y.dep, x=x.indep, training_frame = train.h2o, ntrees = 500, max_depth = 4, learn_rate = 0.01, seed = 1122,distribution="multinomial")
  |                                                                 |   0%
  |===                                                              |   5%
  |=====                                                            |   7%
  |======                                                           |   9%
  |=======                                                          |  10%
  |======================                                           |  33%
  |=====================================                            |  56%
  |====================================================             |  79%
  |================================================================ |  98%
  |=================================================================| 100%
## H2OBinomialMetrics: gbm
## MSE:  0.09860778
## RMSE:  0.3140188
## LogLoss:  0.3206876
## Mean Per-Class Error:  0.2120263
## AUC:  0.8630573
## Gini:  0.7261146
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
##          0  1    Error     Rate
## 0      266 48 0.152866  =48/314
## 1       16 43 0.271186   =16/59
## Totals 282 91 0.171582  =64/373
## Maximum Metrics: Maximum metrics at their respective thresholds
##                       metric threshold    value idx
## 1                     max f1  0.189757 0.573333  90
## 2                     max f2  0.130895 0.693717 145
## 3               max f0point5  0.327346 0.598802  26
## 4               max accuracy  0.442757 0.876676  14
## 5              max precision  0.802184 1.000000   0
## 6                 max recall  0.049990 1.000000 284
## 7            max specificity  0.802184 1.000000   0
## 8           max absolute_mcc  0.169135 0.496486 104
## 9 max min_per_class_accuracy  0.169135 0.796610 104
## 10 max mean_per_class_accuracy  0.169135 0.805948 104
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or `
## Warning in h2o.find_row_by_threshold(object, t): Could not find exact
## threshold: 0.5 for this set of metrics; using closest threshold found:
## 0.501205344484314. Run `h2o.predict` and apply your desired threshold on a
## probability column.
## [[1]]
## [1] 0.1355932
## [1] 0.8630573

This model correctly predicts 43 top 10 hits, which is 10 more than the number predicted by Model 3. Moreover, the AUC metric is higher than the one obtained from Model 3.

As seen above, H2O’s API provides the ability to obtain key statistical measures required to analyze the models easily, using several built-in functions. The record label can experiment with different parameters to arrive at the model that predicts the maximum number of Top 10 hits at the desired level of accuracy and threshold.

H2O also allows you to experiment with deep learning models. Deep learning models have the ability to learn features implicitly, but can be more expensive computationally.

Now, create a deep learning model with the h2o.deeplearning function, using the same training and test datasets created before. The time taken to run this model depends on the type of EC2 instance chosen for this purpose.  For models that require more computation, consider using accelerated computing instances such as the P2 instance type.

  dlearning.modelh <- h2o.deeplearning(y = y.dep,
                                      x = x.indep,
                                      training_frame = train.h2o,
                                      epoch = 250,
                                      hidden = c(250,250),
                                      activation = "Rectifier",
                                      seed = 1122,
  |                                                                 |   0%
  |===                                                              |   4%
  |=====                                                            |   8%
  |========                                                         |  12%
  |==========                                                       |  16%
  |=============                                                    |  20%
  |================                                                 |  24%
  |==================                                               |  28%
  |=====================                                            |  32%
  |=======================                                          |  36%
  |==========================                                       |  40%
  |=============================                                    |  44%
  |===============================                                  |  48%
  |==================================                               |  52%
  |====================================                             |  56%
  |=======================================                          |  60%
  |==========================================                       |  64%
  |============================================                     |  68%
  |===============================================                  |  72%
  |=================================================                |  76%
  |====================================================             |  80%
  |=======================================================          |  84%
  |=========================================================        |  88%
  |============================================================     |  92%
  |==============================================================   |  96%
  |=================================================================| 100%
##    user  system elapsed 
##   1.216   0.020 166.508
## H2OBinomialMetrics: deeplearning
## MSE:  0.1678359
## RMSE:  0.4096778
## LogLoss:  1.86509
## Mean Per-Class Error:  0.3433013
## AUC:  0.7568822
## Gini:  0.5137644
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
##          0  1    Error     Rate
## 0      290 24 0.076433  =24/314
## 1       36 23 0.610169   =36/59
## Totals 326 47 0.160858  =60/373
## Maximum Metrics: Maximum metrics at their respective thresholds
##                       metric threshold    value idx
## 1                     max f1  0.826267 0.433962  46
## 2                     max f2  0.000000 0.588235 239
## 3               max f0point5  0.999929 0.511811  16
## 4               max accuracy  0.999999 0.865952  10
## 5              max precision  1.000000 1.000000   0
## 6                 max recall  0.000000 1.000000 326
## 7            max specificity  1.000000 1.000000   0
## 8           max absolute_mcc  0.999929 0.363219  16
## 9 max min_per_class_accuracy  0.000004 0.662420 145
## 10 max mean_per_class_accuracy  0.000000 0.685334 224
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or `h2o.gainsLift(<model>, valid=<T/F>, xval=<T/F>)`
## Warning in h2o.find_row_by_threshold(object, t): Could not find exact
## threshold: 0.5 for this set of metrics; using closest threshold found:
## 0.496293348880151. Run `h2o.predict` and apply your desired threshold on a
## probability column.
## [[1]]
## [1] 0.3898305
## [1] 0.7568822

The AUC metric for this model is 0.7568822, which is less than what you got from the earlier models. I recommend further experimentation using different hyper parameters, such as the learning rate, epoch or the number of hidden layers.

H2O’s built-in functions provide many key statistical measures that can help measure model performance. Here are some of these key terms.

Metric Description
Sensitivity Measures the proportion of positives that have been correctly identified. It is also called the true positive rate, or recall.
Specificity Measures the proportion of negatives that have been correctly identified. It is also called the true negative rate.
Threshold Cutoff point that maximizes specificity and sensitivity. While the model may not provide the highest prediction at this point, it would not be biased towards positives or negatives.
Precision The fraction of the documents retrieved that are relevant to the information needed, for example, how many of the positively classified are relevant

Provides insight into how well the classifier is able to separate the two classes. The implicit goal is to deal with situations where the sample distribution is highly skewed, with a tendency to overfit to a single class.

0.90 – 1 = excellent (A)

0.8 – 0.9 = good (B)

0.7 – 0.8 = fair (C)

.6 – 0.7 = poor (D)

0.5 – 0.5 = fail (F)

Here’s a summary of the metrics generated from H2O’s built-in functions for the three models that produced useful results.

Metric Model 3 GBM Model Deep Learning Model



















1.0 1.0





1.0 1.0





0.2033898 0.1355932



AUC 0.8492389 0.8630573 0.756882

Note: ‘t’ denotes threshold.

Your options at this point could be narrowed down to Model 3 and the GBM model, based on the AUC and accuracy metrics observed earlier.  If the slightly lower accuracy of the GBM model is deemed acceptable, the record label can choose to go to production with the GBM model, as it can predict a higher number of Top 10 hits.  The AUC metric for the GBM model is also higher than that of Model 3.

Record labels can experiment with different learning techniques and parameters before arriving at a model that proves to be the best fit for their business. Because deep learning models can be computationally expensive, record labels can choose more powerful EC2 instances on AWS to run their experiments faster.


In this post, I showed how the popular music industry can use analytics to predict the type of songs that make the Top 10 Billboard charts. By running H2O’s scalable machine learning platform on AWS, data scientists can easily experiment with multiple modeling techniques and interactively query the data using Amazon Athena, without having to manage the underlying infrastructure. This helps record labels make critical decisions on the type of artists and songs to promote in a timely fashion, thereby increasing sales and revenue.

If you have questions or suggestions, please comment below.

Additional Reading

Learn how to build and explore a simple geospita simple GEOINT application using SparkR.

About the Authors

gopalGopal Wunnava is a Partner Solution Architect with the AWS GSI Team. He works with partners and customers on big data engagements, and is passionate about building analytical solutions that drive business capabilities and decision making. In his spare time, he loves all things sports and movies related and is fond of old classics like Asterix, Obelix comics and Hitchcock movies.



Bob Strahan, a Senior Consultant with AWS Professional Services, contributed to this post.



timeShift(GrafanaBuzz, 1w) Issue 17

Post Syndicated from Blogs on Grafana Labs Blog original https://grafana.com/blog/2017/10/13/timeshiftgrafanabuzz-1w-issue-17/

It’s been a busy week here at Grafana Labs. While we’ve been working on GrafanaCon EU preparations here at the NYC office, the Stockholm office has been diligently working to release Grafana 4.6-beta-1. We’re really excited about this latest release and look forward to your feedback on the new features.

Latest Release

Grafana 4.6-beta-1 is now available! Grafana v4.6 brings many enhancements to Annotations, Cloudwatch and Prometheus. It also adds support for Postgres as a metric and table data source!

To see more details on what’s in the newest version, please see the release notes.

Download Grafana 4.6.0-beta-1 Now

From the Blogosphere

Using Kafka and Grafana to Monitor Meteorological Conditions: Oliver was looking for a way to track historical mountain conditions around the UK, but only had available data for the last 24 hours. It seemed like a perfect job for Kafka. This post discusses how to get going with Kafka very easily, store the data in Graphite and visualize the data in Grafana.

Web Interfaces for your Syslog Server – An Overview: System administrators often prefer to use the command line, but complex queries can be completed much faster with logs indexed in a database and a web interface. This article provides a run-down of various GUI-based tools available for your syslog server.

JEE Performance with JMeter, Prometheus and Grafana. Complete Project from Scratch: This comprehensive article walks you through the steps of monitoring JEE application performance from scratch. We start with making implementation decisions, then how to collect data, visualization and dashboarding configuration, and conclude with alerting. Buckle up; it’s a long article, with a ton of information.

Early Bird Tickets Now Available

Early bird tickets are going fast, so take advantage of the discounted price before they’re gone! We will be announcing the first block of speakers in the coming week.

There’s still time to submit a talk. We’ll accept submissions through the end of October. We’re accepting technical and non-technical talks of all sizes. Submit a CFP.

Get Your Early Bird Ticket Now

Grafana Plugins

This week we add the Prometheus Alertmanager Data Source to our growing list of plugins, lots of updates to the GLPI Data source, and have a urgent bugfix for the WorldMap Panel. To update plugins from on-prem Grafana, use the Grafana-cli tool, or with 1 click if you are using Hosted Grafana.


Prometheus Alertmanager Data Source – This new data source lets you show data from the Prometheus Alertmanager in Grafana. The Alertmanager handles alerts sent by client applications such as the Prometheus server. With this data source, you can show data in Table form or as a SingleStat.

Install Now


WorldMap Panel – A new version with an urgent bugfix for Elasticsearch users:

  • A fix for Geohash maps after a breaking change in Grafana 4.5.0.
  • Last Geohash as center for the map – it centers the map on the last geohash position received. Useful for real time tracking (with auto refresh on in Grafana).



GLPI App – Lots of fixes in the new version:

  • Compatibility with GLPI 9.2
  • Autofill the Timerange field based on the query
  • When adding new query, add by default a ticket query instead of undefined
  • Correct values in hover tooltip
  • Can have element count by hour of the day with the panel histogram


Contributions of the week:

Each week we highlight some of the important contributions from our amazing open source community. Thank you for helping make Grafana better!

Grafana Labs is Hiring!

We are passionate about open source software and thrive on tackling complex challenges to build the future. We ship code from every corner of the globe and love working with the community. If this sounds exciting, you’re in luck – WE’RE HIRING!

Check out our Open Positions

New Annotation Function

In addition to being able to add annotations easily in the graph panel, you can also create ranges as shown above. Give 4.6.0-beta-1 a try and give us your feedback.

We Need Your Help!

Do you have a graph that you love because the data is beautiful or because the graph provides interesting information? Please get in touch. Tweet or send us an email with a screenshot, and we’ll tell you about this fun experiment.

Tell Me More

What do you think?

We want to keep these articles interesting and relevant, so please tell us how we’re doing. Submit a comment on this article below, or post something at our community forum. Help us make these weekly roundups better!

Follow us on Twitter, like us on Facebook, and join the Grafana Labs community.

Popcorn Time Creator Readies BitTorrent & Blockchain-Powered Video Platform

Post Syndicated from Andy original https://torrentfreak.com/popcorn-time-creator-readies-bittorrent-blockchain-powered-youtube-competitor-171012/

Without a doubt, YouTube is one of the most important websites available on the Internet today.

Its massive archive of videos brings pleasure to millions on a daily basis but its centralized nature means that owner Google always exercises control.

Over the years, people have looked to decentralize the YouTube concept and the latest project hoping to shake up the market has a particularly interesting player onboard.

Until 2015, only insiders knew that Argentinian designer Federico Abad was actually ‘Sebastian’, the shadowy figure behind notorious content sharing platform Popcorn Time.

Now he’s part of the team behind Flixxo, a BitTorrent and blockchain-powered startup hoping to wrestle a share of the video market from YouTube. Here’s how the team, which features blockchain startup RSK Labs, hope things will play out.

The Flixxo network will have no centralized storage of data, eliminating the need for expensive hosting along with associated costs. Instead, transfers will take place between peers using BitTorrent, meaning video content will be stored on the machines of Flixxo users. In practice, the content will be downloaded and uploaded in much the same way as users do on The Pirate Bay or indeed Abad’s baby, Popcorn Time.

However, there’s a twist to the system that envisions content creators, content consumers, and network participants (seeders) making revenue from their efforts.

At the heart of the Flixxo system are digital tokens (think virtual currency), called Flixx. These Flixx ‘coins’, which will go on sale in 12 days, can be used to buy access to content. Creators can also opt to pay consumers when those people help to distribute their content to others.

“Free from structural costs, producers can share the earnings from their content with the network that supports them,” the team explains.

“This way you get paid for helping us improve Flixxo, and you earn credits (in the form of digital tokens called Flixx) for watching higher quality content. Having no intermediaries means that the price you pay for watching the content that you actually want to watch is lower and fairer.”

The Flixxo team

In addition to earning tokens from helping to distribute content, people in the Flixxo ecosystem can also earn currency by watching sponsored content, i.e advertisements. While in a traditional system adverts are often considered a nuisance, Flixx tokens have real value, with a promise that users will be able to trade their Flixx not only for videos, but also for tangible and semi-tangible goods.

“Use your Flixx to reward the producers you follow, encouraging them to create more awesome content. Or keep your Flixx in your wallet and use them to buy a movie ticket, a pair of shoes from an online retailer, a chest of coins in your favourite game or even convert them to old-fashioned cash or up-and-coming digital assets, like Bitcoin,” the team explains.

The Flixxo team have big plans. After foundation in early 2016, the second quarter of 2017 saw the completion of a functional alpha release. In a little under two weeks, the project will begin its token generation event, with new offices in Los Angeles planned for the first half of 2018 alongside a premiere of the Flixxo platform.

“A total of 1,000,000,000 (one billion) Flixx tokens will be issued. A maximum of 300,000,000 (three hundred million) tokens will be sold. Some of these tokens (not more than 33% or 100,000,000 Flixx) may be sold with anticipation of the token allocation event to strategic investors,” Flixxo states.

Like all content platforms, Flixxo will live or die by the quality of the content it provides and whether, at least in the first instance, it can persuade people to part with their hard-earned cash. Only time will tell whether its content will be worth a premium over readily accessible YouTube content but with much-reduced costs, it may tempt creators seeking a bigger piece of the pie.

“Flixxo will also educate its community, teaching its users that in this new internet era value can be held and transferred online without intermediaries, a value that can be earned back by participating in a community, by contributing, being rewarded for every single social interaction,” the team explains.

Of course, the elephant in the room is what will happen when people begin sharing copyrighted content via Flixxo. Certainly, the fact that Popcorn Time’s founder is a key player and rival streaming platform Stremio is listed as a partner means that things could get a bit spicy later on.

Nevertheless, the team suggests that piracy and spam content distribution will be limited by mechanisms already built into the system.

“[A]uthors have to time-block tokens in a smart contract (set as a warranty) in order to upload content. This contract will also handle and block their earnings for a certain period of time, so that in the case of a dispute the unfair-uploader may lose those tokens,” they explain.

That being said, Flixxo also says that “there is no way” for third parties to censor content “which means that anyone has the chance of making any piece of media available on the network.” However, Flixxo says it will develop tools for filtering what it describes as “inappropriate content.”

At this point, things start to become a little unclear. On the one hand Flixxo says it could become a “revolutionary tool for uncensorable and untraceable media” yet on the other it says that it’s necessary to ensure that adult content, for example, isn’t seen by kids.

“We know there is a thin line between filtering or curating content and censorship, and it is a fact that we have an open network for everyone to upload any content. However, Flixxo as a platform will apply certain filtering based on clear rules – there should be a behavior-code for uploaders in order to offer the right content to the right user,” Flixxo explains.

To this end, Flixxo says it will deploy a centralized curation function, carried out by 101 delegates elected by the community, which will become progressively decentralized over time.

“This curation will have a cost, paid in Flixx, and will be collected from the warranty blocked by the content uploaders,” they add.

There can be little doubt that if Flixxo begins ‘curating’ unsuitable content, copyright holders will call on it to do the same for their content too. And, if the platform really takes off, 101 curators probably won’t scratch the surface. There’s also the not inconsiderable issue of what might happen to curators’ judgment when they’re incentivized to block curate content.

Finally, for those sick of “not available in your region” messages, there’s good and bad news. Flixxo insists there will be no geo-blocking of content on its part but individual creators will still have that feature available to them, should they choose.

The Flixx whitepaper can be downloaded here (pdf)

Source: TF, for the latest info on copyright, file-sharing, torrent sites and ANONYMOUS VPN services.