Mailing-List: contact general-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: general@incubator.apache.org
Received-SPF: neutral (nike.apache.org: local policy)
Message-Id: <3BEA62CE-B2D6-4474-B76E-FE176A9D2BFE@yahoo-inc.com>
From: Alan Gates <gates@yahoo-inc.com>
To: general@incubator.apache.org
Content-Type: text/plain; charset=WINDOWS-1252; format=flowed; delsp=yes
Content-Transfer-Encoding: quoted-printable
Subject: [VOTE] Accept Howl as an Incubator Project
Mime-Version: 1.0 (Apple Message framework v936)
Date: Tue, 22 Feb 2011 16:20:13 -0800

I would like to call a vote on accepting Howl as an Incubator =20
project.  The proposal is available at =
http://wiki.apache.org/incubator/HowlProposal=20
.  You can see the discussion from the proposal thread at =
http://tinyurl.com/5w7y9p9=20
.

Alan.

----------------------

Abstract
Howl is a table and storage management service for data created using =20=

Apache Hadoop.


Proposal
The vision of Howl is to provide table management and storage =20
management layers for Apache Hadoop. This includes:

	=95 Providing a shared schema and data type mechanism.
	=95 Providing a table abstraction so that users need not be =
concerned =20
with where or how their data is stored.
	=95 Providing interoperability across data processing tools such =
as =20
Pig, Map Reduce, Streaming, and Hive.

Background
Data processors using Apache Hadoop have a common need for table =20
management services. The goal of a table management service is to =20
track data that exists in a Hadoop grid and present that data to users =20=

in a tabular format. Such a table management service needs to provide =20=

a single input and output format to users so that individual users =20
need not be concerned with the storage formats that are chosen for =20
particular data sets. As part of having a single format, the data will =20=

need to be described by one type of schema and have a single datatype =20=

system.

Additionally, users should be free to choose the best tools for their =20=

use cases. The Hadoop project includes Map Reduce, Streaming, Pig, and =20=

Hive, and additional tools exist such as Cascading. Each of these =20
tools has users who prefer it, and there are use cases best addressed =20=

by each of these tools. Two users on the same grid who need to share =20
data should not be constrained to use the same tool but rather should =20=

be free to choose the best tool for their use case. A table management =20=

service that presents data in the same way to all of the tools can =20
alleviate this problem by providing interfaces to each of the data =20
processing tools.

There are also a few other features a table management service should =20=

provide, such as notification of when data arrives.

A couple of developers at Yahoo! started the project. It is based on =20
the Hive MetaStore component. There is good amount of interest in such =20=

a service expressed from Yahoo!, Facebook, LinkedIn, and, others. We =20
are therefore proposing to place Howl in the Apache incubator and to =20
build an open source community around it.


Rationale
There is a strong need for a table management service, especially for =20=

large grids with petabytes of data, and where the data volume is =20
increasing by the day. Hadoop users need to find data to read and have =20=

a place to store  their data. Currently users must understand the =20
location of data to read, the storage format, compression techniques =20
used, etc. To write data they need to understand where on HDFS their =20
data belongs, the best compression format to use, how their data =20
should be serialized, etc.

Most users do not want to be concerned with these issues. They want =20
these managed for them.

Having it as an Apache Open Source project will highly benefit Howl =20
from the point of view of getting a large community that currently =20
uses Hadoop and the other products built around Hadoop (like Pig, =20
Hive, etc.). Users of the Hadoop ecosystem can influence Howl=92s =20
roadmap, and contribute to it. Looking at it in another way, we =20
believe having Howl as part of the Hadoop ecosystem will be a great =20
benefit to the current Hadoop/Pig/Hive community too.


Current Status

Meritocracy
Our intent with this incubator proposal is to start building a diverse =20=

developer community around Howl following the Apache meritocracy =20
model. We have wanted to make the project open source and encourage =20
contributors from multiple organizations from the start. We plan to =20
provide plenty of support to new developers and to quickly recruit =20
those who make solid contributions to committer status.


Community
Howl is currently being used by developers at Yahoo! and there has =20
been an expressed interest from LinkedIn and Facebook. Yahoo! also =20
plans to deploy the current version of Howl in production soon. We =20
hope to extend the user and developer base further in the future. The =20=

current developers and users are all interested in building a solid =20
open source community around Howl.

To work towards an open source community, we have started using the =20
GitHub issue tracker and mailing lists at Yahoo! for development =20
discussions within our group.


Core Developers
Howl is currently being developed by four engineers from Yahoo! - =20
Devaraj Das, Ashutosh Chauhan, Sushanth Sowmyan, and Mac Yang. All the =20=

engineers have deep expertise in Hadoop and the Hadoop Ecosystem in =20
general.


Alignment
The ASF is a natural host for Howl given that it is already the home =20
of Hadoop, Pig, HBase, Cassandra, and other emerging cloud software =20
projects. Howl was designed to support Hadoop from the beginning in =20
order to solve data management challenges in Hadoop clusters. Howl =20
complements the existing Apache cloud computing projects by providing =20=

a unified way to manage data.


Known Risks

Orphaned Products
The core developers plan to work full time on the project. There is =20
very little risk of Howl getting orphaned since large companies like =20
Yahoo! are planning to deploy this in their production Hadoop =20
clusters. We believe we can build an active developer community around =20=

Howl (companies like Facebook and LinkedIn have also expressed =20
interest).


Inexperience with Open Source
All of the core developers are active users and followers of open =20
source. Devaraj Das is an Apache Hadoop committer and Apache Hadoop =20
PMC member, and has experience with the Apache infrastructure and =20
development process. Ashutosh Chauhan is an Apache Pig committer and =20
Apache Pig PMC member. Sushanth Sowmyan and Mac Yang made =20
contributions to the Apache Hive and the Apache Chukwa projects.


Homogeneous Developers
The current core developers are all from Yahoo! However, we hope to =20
establish a developer community that includes contributors from =20
several corporations, and we are starting to work towards this with =20
Facebook and LinkedIn.


Reliance on Salaried Developers
Currently, the developers are paid to do work on Howl. However, once =20
the project has a community built around it, we expect to get =20
committers and developers from outside the current core developers. =20
Companies like Yahoo! are invested in Howl being a solution to the =20
data management problem in Hadoop clusters, and that is not likely to =20=

change.


Relationships with Other Apache Products
Howl is going to be used by users of Hadoop, Pig, and Hive. See =20
section Initial Source below for more information about Howl's =20
relationship to Hive.


An Excessive Fascination with the Apache Brand
While we respect the reputation of the Apache brand and have no doubts =20=

that it will attract contributors and users, our interest is primarily =20=

to give Howl a solid home as an open source project following an =20
established development model. We have also given reasons in the =20
Rationale and Alignment sections.


Documentation
Information about Howl can be found at http://wiki.apache.org/pig/=20
Howl. The following sources may be useful to start with:

	=95
The GitHub site: https://github.com/yahoo/howl

	=95
The roadmap: http://wiki.apache.org/pig/HowlJournal


Initial Source
Howl has been under development since Summer 2010 by a team of =20
engineers in Yahoo!. It is currently hosted on GitHub under an Apache =20=

license at https://github.com/yahoo/howl.

The initial development of Howl has consisted of:

	=95 maintaining a branch of the entire Hive codebase
	=95 getting Howl-related patches committed to Hive
	=95 developing Howl-specific plugins and wrappers to customize =
Hive =20
behavior
At runtime, Howl executes Hive code for metastore and CLI+DDL, =20
disabling anything related to Hadoop map/reduce execution. It also =20
makes use of the RCFile storage format contained in Hive.

This approach was taken as a first step in order to validate the =20
required functionality and get a production version working. However, =20=

in the long-term, maintaining a clone of Hive is undesirable. One =20
possible resolution is to factor the metastore+CLI+DDL components out =20=

of Hive and move them into Howl (making Hive dependent on Howl). =20
Another possible resolution is to remove the copy of Hive from Howl =20
and do the build/release engineering necessary to make Howl depend on =20=

Hive. As part of the incubation process, we plan to work towards =20
resolution of  these issues.


External Dependencies
The dependencies all have Apache compatible licenses.


Cryptography
Not applicable.


Required Resources

Mailing Lists
	=95 howl-private for private PMC discussions (with moderated =20
subscriptions)
	=95 howl-dev
	=95 howl-commits
	=95 howl-user

Subversion Directory
https://svn.apache.org/repos/asf/incubator/howl


Issue Tracking
JIRA Howl (HOWL)


Other Resources
The existing code already has unit tests, so we would like a Hudson =20
instance to run them whenever a new patch is submitted. This can be =20
added after project creation.


Initial Committers
	=95 Devaraj Das
	=95 Ashutosh Chauhan
	=95 Sushanth Sowmyan
	=95 Mac Yang
	=95 Paul Yang
	=95 Alan Gates
A CLA is already on file for Sushanth.


Affiliations
	=95 Devaraj Das (Yahoo!)
	=95 Ashutosh Chauhan (Yahoo!)
	=95 Sushanth Sowmyan (Yahoo!)
	=95 Mac Yang (Yahoo!)
	=95 Paul Yang (Facebook)
	=95 Alan Gates (Yahoo!)

Sponsors

Champion
Owen O=92Malley


Nominated Mentors
	=95 Olga Natkovich (Pig PMC member and Apache VP for Pig)
	=95 Alan Gates (Pig PMC member)
	=95 John Sichi (Hive PMC member)

Sponsoring Entity
We are requesting the Incubator to sponsor this project.=

---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@incubator.apache.org
For additional commands, e-mail: general-help@incubator.apache.org