incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Manoharan, Arun" <>
Subject Re: [DISCUSS] Eagle incubator proposal
Date Tue, 20 Oct 2015 23:14:13 GMT
Thanks Taylor. I will add you to the mentor list.

On 10/20/15, 11:58 AM, "P. Taylor Goetz" <> wrote:

>I should also have some improved bandwidth both now that Kylin is nearing
>graduation and for other reasons. I¹ve been bogged down recently, but
>that¹s starting to change.
>If more mentors are desired, I¹d be willing to help in that respect.
>> On Oct 20, 2015, at 11:49 AM, Henry Saputra <>
>> Hi Ted,
>> Since Kylin almost ready to graduate, I have more bandwidth to help
>>with Eagle.
>> But, you are right that current proposed mentors for Eagle seemed to
>> be very busy with other podlings, so 1 or 2 additional mentors would
>> be great.
>> The good news is that the team consist some people from Kylin, for
>> example Luke, which done great job helping Kylin to understand working
>> with Apache way.
>> So we have some help from initial committers who have done the rodeo
>> - Henry
>> On Mon, Oct 19, 2015 at 9:00 AM, Ted Dunning <>
>>> I would suggest that Owen O'Malley has not had enough time to be a
>>> mentor recently and should not be on the list of mentors.
>>> Henry and Julian are good if their schedules permit.  Henry, I know has
>>> been mentoring a number of projects lately.
>>> On Mon, Oct 19, 2015 at 8:40 AM, Jean-Baptiste Onofré <>
>>> wrote:
>>>> Hi Arun,
>>>> very interesting proposal. I may see some possible interaction with
>>>> Falcon. In Falcon, we have HDFS files (and Hive/HBase) monitoring
>>>>(with a
>>>> kind of Change Data Capture), etc.
>>>> So, I see a different perspective in Eagle, but Eagle could also
>>>> Falcon somehow.
>>>> Regards
>>>> JB
>>>> On 10/19/2015 05:33 PM, Manoharan, Arun wrote:
>>>>> Hello Everyone,
>>>>> My name is Arun Manoharan. Currently a product manager in the
>>>>> platform team at eBay Inc.
>>>>> I would like to start a discussion on Eagle and its joining the ASF
>>>>>as an
>>>>> incubation project.
>>>>> Eagle is a Monitoring solution for Hadoop to instantly identify
>>>>>access to
>>>>> sensitive data, recognize attacks, malicious activities and take
>>>>>actions in
>>>>> real time. Eagle supports a wide variety of policies on HDFS data
>>>>>and Hive.
>>>>> Eagle also provides machine learning models for detecting anomalous
>>>>> behavior in Hadoop.
>>>>> The proposal is available on the wiki here:
>>>>> The text of the proposal is also available at the end of this email.
>>>>> Thanks for your time and help.
>>>>> Thanks,
>>>>> Arun
>>>>> <COPY of the proposal in text format>
>>>>> Eagle
>>>>> Abstract
>>>>> Eagle is an Open Source Monitoring solution for Hadoop to instantly
>>>>> identify access to sensitive data, recognize attacks, malicious
>>>>> in hadoop and take actions.
>>>>> Proposal
>>>>> Eagle audits access to HDFS files, Hive and HBase tables in real
>>>>> enforces policies defined on sensitive data access and alerts or
>>>>> user¹s access to that sensitive data in real time. Eagle also
>>>>>creates user
>>>>> profiles based on the typical access behaviour for HDFS and Hive and
>>>>> alerts when anomalous behaviour is detected. Eagle can also import
>>>>> sensitive data information classified by external classification
>>>>>engines to
>>>>> help define its policies.
>>>>> Overview of Eagle
>>>>> Eagle has 3 main parts.
>>>>> 1.Data collection and storage - Eagle collects data from various
>>>>> logs in real time using Kafka/Yarn API and uses HDFS and HBase for
>>>>> 2.Data processing and policy engine - Eagle allows users to create
>>>>> policies based on various metadata properties on HDFS, Hive and
>>>>>HBase data.
>>>>> 3.Eagle services - Eagle services include policy manager, query
>>>>> and the visualization component. Eagle provides intuitive user
>>>>>interface to
>>>>> administer Eagle and an alert dashboard to respond to real time
>>>>> Data Collection and Storage:
>>>>> Eagle provides programming API for extending Eagle to integrate any
>>>>> source into Eagle policy evaluation framework. For example, Eagle
>>>>> audit monitoring collects data from Kafka which is populated from
>>>>> log4j appender or from logstash agent. Eagle hive monitoring
>>>>>collects hive
>>>>> query logs from running job through YARN API, which is designed to be
>>>>> scalable and fault-tolerant. Eagle uses HBase as storage for storing
>>>>> metadata and metrics data, and also supports relational database
>>>>> configuration change.
>>>>> Data Processing and Policy Engine:
>>>>> Processing Engine: Eagle provides stream processing API which is an
>>>>> abstraction of Apache Storm. It can also be extended to other
>>>>> engines. This abstraction allows developers to assemble data
>>>>> transformation, filtering, external data join etc. without
>>>>>physically bound
>>>>> to a specific streaming platform. Eagle streaming API allows
>>>>>developers to
>>>>> easily integrate business logic with Eagle policy engine and
>>>>> Eagle framework compiles business logic execution DAG into program
>>>>> primitives of underlying stream infrastructure e.g. Apache Storm. For
>>>>> example, Eagle HDFS monitoring transforms audit log from Namenode to
>>>>> and joins sensitivity metadata, security zone metadata which are
>>>>> from external programs or configured by user. Eagle hive monitoring
>>>>> running jobs to get hive query string and parses query string into
>>>>> and then joins sensitivity metadata.
>>>>> Alerting Framework: Eagle Alert Framework includes stream metadata
>>>>> scalable policy engine framework, extensible policy engine framework.
>>>>> Stream metadata API allows developers to declare event schema
>>>>> what attributes constitute an event, what is the type for each
>>>>> and how to dynamically resolve attribute value in runtime when user
>>>>> configures policy. Scalable policy engine framework allows policies
>>>>>to be
>>>>> executed on different physical nodes in parallel. It is also used to
>>>>> your own policy partitioner class. Policy engine framework together
>>>>> streaming partitioning capability provided by all streaming
>>>>>platforms will
>>>>> make sure policies and events can be evaluated in a fully
>>>>>distributed way.
>>>>> Extensible policy engine framework allows developer to plugin a new
>>>>> engine with a few lines of codes. WSO2 Siddhi CEP engine is the
>>>>> engine which Eagle supports as first-class citizen.
>>>>> Machine Learning module: Eagle provides capabilities to define user
>>>>> activity patterns or user profiles for Hadoop users based on the user
>>>>> behaviour in the platform. These user profiles are modeled using
>>>>> Learning algorithms and used for detection of anomalous users
>>>>> Eagle uses Eigen Value Decomposition, and Density Estimation
>>>>>algorithms for
>>>>> generating user profile models. The model reads data from HDFS audit
>>>>> preprocesses and aggregates data, and generates models using Spark
>>>>> programming APIs. Once models are generated, Eagle uses stream
>>>>> engine for near real-time anomaly detection to determine if any
>>>>> activities are suspicious or not.
>>>>> Eagle Services:
>>>>> Query Service: Eagle provides SQL-like service API to support
>>>>> comprehensive computation for huge set of data on the fly, for e.g.
>>>>> comprehensive filtering, aggregation, histogram, sorting, top,
>>>>> expression, pagination etc. HBase is the data storage which Eagle
>>>>> as first-class citizen, relational database is supported as well.
>>>>>For HBase
>>>>> storage, Eagle query framework compiles user provided SQL-like query
>>>>> HBase native filter objects and execute it through HBase coprocessor
>>>>>on the
>>>>> fly.
>>>>> Policy Manager: Eagle policy manager provides UI and Restful API for
>>>>> to define policy with just a few clicks. It includes site management
>>>>> policy editor, sensitivity metadata import, HDFS or Hive sensitive
>>>>> browsing, alert dashboards etc.
>>>>> Background
>>>>> Data is one of the most important assets for today¹s businesses,
>>>>> makes data security one of the top priorities of today¹s enterprises.
>>>>> Hadoop is widely used across different verticals as a big data
>>>>> to store this data in most modern enterprises.
>>>>> At eBay we use hadoop platform extensively for our data processing
>>>>> Our data in Hadoop is becoming bigger and bigger as our user base is
>>>>> an exponential growth. Today there are variety of data sets
>>>>>available in
>>>>> Hadoop cluster for our users to consume. eBay has around 120 PB of
>>>>> stored in HDFS across 6 different clusters and around 1800+ active
>>>>> users consuming data thru Hive, HBase and mapreduce jobs everyday to
>>>>> applications using this data. With this astronomical growth of data
>>>>> are also challenges in securing sensitive data and monitoring the
>>>>>access to
>>>>> this sensitive data. Today in large organizations HDFS is the defacto
>>>>> standard for storing big data. Data sets which includes and not
>>>>>limited to
>>>>> consumer sentiment, social media data, customer segmentation, web
>>>>> sensor data, geo-location and transaction data get stored in Hadoop
>>>>>for day
>>>>> to day business needs.
>>>>> We at eBay want to make sure the sensitive data and data platforms
>>>>> completely protected from security breaches. So we partnered very
>>>>> with our Information Security team to understand the requirements
>>>>>for Eagle
>>>>> to monitor sensitive data access on hadoop:
>>>>> 1.Ability to identify and stop security threats in real time
>>>>> 2.Scale for big data (Support PB scale and Billions of events)
>>>>> 3.Ability to create data access policies
>>>>> 4.Support multiple data sources like HDFS, HBase, Hive
>>>>> 5.Visualize alerts in real time
>>>>> 6.Ability to block malicious access in real time
>>>>> We did not find any data access monitoring solution that available
>>>>> and can provide the features and functionality that we need to
>>>>>monitor the
>>>>> data access in the hadoop ecosystem at our scale. Hence with an
>>>>> team of world class developers and several users, we have been able
>>>>> bring Eagle into production as well as open source it.
>>>>> Rationale
>>>>> In today¹s world; data is an important asset for any company.
>>>>> are using data extensively to create amazing experiences for users.
>>>>> has to be protected and access to data should be secured from
>>>>> breaches. Today Hadoop is not only used to store logs but also stores
>>>>> financial data, sensitive data sets, geographical data, user click
>>>>> data sets etc. which makes it more important to be protected from
>>>>> breaches. To secure a data platform there are multiple things that
>>>>>need to
>>>>> happen. One is having a strong access control mechanism which today
>>>>> provided by Apache Ranger and Apache Sentry. These tools provide the
>>>>> ability to provide fine grain access control mechanism to data sets
>>>>> hadoop. But there is a big gap in terms of monitoring all the data
>>>>> events and activities in order to securing the hadoop data platform.
>>>>> Together with strong access control, perimeter security and data
>>>>> monitoring in place data in the hadoop clusters can be secu
>>>> r
>>>> ed against breaches. We looked around and found following:
>>>>> Existing data activity monitoring products are designed for
>>>>> databases and data warehouse. Existing monitoring platforms cannot
>>>>> out to support fast growing data and petabyte scale. Few products in
>>>>> industry are still very early in terms of supporting HDFS, Hive,
>>>>>HBase data
>>>>> access monitoring.
>>>>> As mentioned in the background, the business requirement and urgency
>>>>> secure the data from users with malicious intent drove eBay to
>>>>>invest in
>>>>> building a real time data access monitoring solution from scratch to
>>>>> real time alerts and remediation features for malicious data access.
>>>>> With the power of open source distributed systems like Hadoop, Kafka
>>>>> much more we were able to develop a data activity monitoring system
>>>>> can scale, identify and stop malicious access in real time.
>>>>> Eagle allows admins to create standard access policies and rules for
>>>>> monitoring HDFS, Hive and HBase data. Eagle also provides out of box
>>>>> machine learning models for modeling user profiles based on user
>>>>> behaviour and use the model to alert on anomalies.
>>>>> Current Status
>>>>> Meritocracy
>>>>> Eagle has been deployed in production at eBay for monitoring
>>>>>billions of
>>>>> events per day from HDFS and Hive operations. From the start; the
>>>>> has been built with focus on high scalability and application
>>>>> in mind and Eagle has demonstrated great performance in responding to
>>>>> suspicious events instantly and great flexibility in defining policy.
>>>>> Community
>>>>> Eagle seeks to develop the developer and user communities during
>>>>> incubation.
>>>>> Core Developers
>>>>> Eagle is currently being designed and developed by engineers from
>>>>> Inc. ­ Edward Zhang, Hao Chen, Chaitali Gupta, Libin Sun, Jilin
>>>>> Qingwen Zhao, Senthil Kumar, Hemanth Dendukuri, Arun Manoharan. All
>>>>> these core developers have deep expertise in developing monitoring
>>>>> for the Hadoop ecosystem.
>>>>> Alignment
>>>>> The ASF is a natural host for Eagle given that it is already the
>>>>>home of
>>>>> Hadoop, HBase, Hive, Storm, Kafka, Spark and other emerging big data
>>>>> projects. Eagle leverages lot of Apache open-source products. Eagle
>>>>> designed to offer real time insights into sensitive data access by
>>>>> monitoring the data access on various data sets in hadoop and an
>>>>> alerting framework with a powerful policy engine. Eagle compliments
>>>>> existing Hadoop platform area by providing a comprehensive
>>>>>monitoring and
>>>>> alerting solution for detecting sensitive data access threats based
>>>>> preset policies and machine learning models for user behaviour
>>>>> Known Risks
>>>>> Orphaned Products
>>>>> The core developers of Eagle team work full time on this project.
>>>>> is no risk of Eagle getting orphaned since eBay is extensively using
>>>>>it in
>>>>> their production Hadoop clusters and have plans to go beyond hadoop.
>>>>> example, currently there are 7 hadoop clusters and 2 of them are
>>>>> monitored using Hadoop Eagle in production. We have plans to extend
>>>>>it to
>>>>> all hadoop clusters and eventually other data platforms. There are
>>>>>10¹s of
>>>>> policies onboarded and actively monitored with plans to onboard more
>>>>> case. We are very confident that every hadoop cluster in the world
>>>>>will be
>>>>> monitored using Eagle for securing the hadoop ecosystem by actively
>>>>> monitoring for data access on sensitive data. We plan to extend and
>>>>> diversify this community further through Apache. We presented Eagle
>>>>>at the
>>>>> hadoop summit in china and garnered interest from different
>>>>>companies who
>>>>> use hadoop extensively.
>>>>> Inexperience with Open Source
>>>>> The core developers are all active users and followers of open
>>>>> They are already committers and contributors to the Eagle Github
>>>>> All have been involved with the source code that has been released
>>>>>under an
>>>>> open source license, and several of them also have experience
>>>>> code in an open source environment. Though the core set of
>>>>>Developers do
>>>>> not have Apache Open Source experience, there are plans to onboard
>>>>> individuals with Apache open source experience on to the project.
>>>>> Kylin PMC members are also in the same ebay organization. We work
>>>>> closely with Apache Ranger committers and are looking forward to find
>>>>> meaningful integrations to improve the security of hadoop platform.
>>>>> Homogenous Developers
>>>>> The core developers are from eBay. Today the problem of monitoring
>>>>> activities to find and stop threats is a universal problem faced by
>>>>>all the
>>>>> businesses. Apache Incubation process encourages an open and diverse
>>>>> meritocratic community. Eagle intends to make every possible effort
>>>>> build a diverse, vibrant and involved community and has already
>>>>> substantial interest from various organizations.
>>>>> Reliance on Salaried Developers
>>>>> eBay invested in Eagle as the monitoring solution for Hadoop
>>>>>clusters and
>>>>> some of its key engineers are working full time on the project. In
>>>>> addition, since there is a growing need for securing sensitive data
>>>>> we need a data activity monitoring solution for Hadoop, we look
>>>>>forward to
>>>>> other Apache developers and researchers to contribute to the project.
>>>>> Additional contributors, including Apache committers have plans to
>>>>> this effort shortly. Also key to addressing the risk associated with
>>>>> relying on Salaried developers from a single entity is to increase
>>>>> diversity of the contributors and actively lobby for Domain experts
>>>>>in the
>>>>> security space to contribute. Eagle intends to do this.
>>>>> Relationships with Other Apache Products
>>>>> Eagle has a strong relationship and dependency with Apache Hadoop,
>>>>> Spark, Kafka and Storm. Being part of Apache¹s Incubation community,
>>>>> help with a closer collaboration among these projects and as well as
>>>>> others. An Excessive Fascination with the Apache Brand Eagle is
>>>>> to enter incubation at Apache in order to help efforts to diversify
>>>>> committer-base, not so much to capitalize on the Apache brand. The
>>>>> project is in production use already inside eBay, but is not
>>>>>expected to be
>>>>> an eBay product for external customers. As such, the Eagle project
>>>>>is not
>>>>> seeking to use the Apache brand as a marketing tool.
>>>>> Documentation
>>>>> Information about Eagle can be found at
>>>>> The following link provide more information about Eagle
>>>>> .
>>>>> Initial Source
>>>>> Eagle has been under development since 2014 by a team of engineers at
>>>>> eBay Inc. It is currently hosted on under an Apache
>>>>>license 2.0
>>>>> at Once in incubation we will be
>>>>> the code base to apache git library.
>>>>> External Dependencies
>>>>> Eagle has the following external dependencies.
>>>>> Basic
>>>>> ?JDK 1.7+
>>>>> ?Scala 2.10.4
>>>>> ?Apache Maven
>>>>> ?JUnit
>>>>> ?Log4j
>>>>> ?Slf4j
>>>>> ?Apache Commons
>>>>> ?Apache Commons Math3
>>>>> ?Jackson
>>>>> ?Siddhi CEP engine
>>>>> Hadoop
>>>>> ?Apache Hadoop
>>>>> ?Apache HBase
>>>>> ?Apache Hive
>>>>> ?Apache Zookeeper
>>>>> ?Apache Curator
>>>>> Apache Spark
>>>>> ?Spark Core Library
>>>>> REST Service
>>>>> ?Jersey
>>>>> Query
>>>>> ?Antlr
>>>>> Stream processing
>>>>> ?Apache Storm
>>>>> ?Apache Kafka
>>>>> Web
>>>>> ?AngularJS
>>>>> ?jQuery
>>>>> ?Bootstrap V3
>>>>> ?Moment JS
>>>>> ?Admin LTE
>>>>> ?html5shiv
>>>>> ?respond
>>>>> ?Fastclick
>>>>> ?Date Range Picker
>>>>> ?Flot JS
>>>>> Cryptography
>>>>> Eagle will eventually support encryption on the wire. This is not
>>>>>one of
>>>>> the initial goals, and we do not expect Eagle to be a controlled
>>>>> item due to the use of encryption. Eagle supports but does not
>>>>>require the
>>>>> Kerberos authentication mechanism to access secured Hadoop services.
>>>>> Required Resources
>>>>> Mailing List
>>>>> ?eagle-private for private PMC discussions
>>>>> ?eagle-dev for developers
>>>>> ?eagle-commits for all commits
>>>>> ?eagle-users for all eagle users
>>>>> Subversion Directory
>>>>> ?Git is the preferred source control system.
>>>>> Issue Tracking
>>>>> ?JIRA Eagle (Eagle)
>>>>> Other Resources
>>>>> The existing code already has unit tests so we will make use of
>>>>> Apache continuous testing infrastructure. The resulting load should
>>>>>not be
>>>>> very large.
>>>>> Initial Committers
>>>>> ?Seshu Adunuthula <sadunuthula at ebay dot com>
>>>>> ?Arun Manoharan <armanoharan at ebay dot com>
>>>>> ?Edward Zhang <yonzhang at ebay dot com>
>>>>> ?Hao Chen <hchen9 at ebay dot com>
>>>>> ?Chaitali Gupta <cgupta at ebay dot com>
>>>>> ?Libin Sun <libsun at ebay dot com>
>>>>> ?Jilin Jiang <jiljiang at ebay dot com>
>>>>> ?Qingwen Zhao <qingwzhao at ebay dot com>
>>>>> ?Hemanth Dendukuri <hdendukuri at ebay dot com>
>>>>> ?Senthil Kumar <senthilkumar at ebay dot com>
>>>>> ?Tan Chen <tanchen at ebay dot com>
>>>>> Affiliations
>>>>> The initial committers are employees of eBay Inc.
>>>>> Sponsors
>>>>> Champion
>>>>> ?Henry Saputra <hsaputra at apache dot org> - Apache IPMC member
>>>>> Nominated Mentors
>>>>> ?Owen O¹Malley < omalley at apache dot org > - Apache IPMC member,
>>>>> Hortonworks
>>>>> ?Henry Saputra <hsaputra at apache dot org> - Apache IPMC member
>>>>> ?Julian Hyde <jhyde at hortonworks dot com> - Apache IPMC member,
>>>>> Hortonworks
>>>>> Sponsoring Entity
>>>>> We are requesting the Incubator to sponsor this project.
>>>> --
>>>> Jean-Baptiste Onofré
>>>> Talend -
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail:
>>>> For additional commands, e-mail:
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail:
>> For additional commands, e-mail:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message