stratos-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sanjiva Weerawarana <>
Subject Fwd: [PROPOSAL] Stratosphere
Date Sun, 30 Mar 2014 21:22:59 GMT
Hmmm will Stratos and Stratosphere be confusing? If so we should give
feedback on the name ...

---------- Forwarded message ----------
From: Alan Gates <>
Date: Sun, Mar 30, 2014 at 11:14 AM
Subject: [PROPOSAL] Stratosphere

I would like to propose Stratosphere as an Apache Incubator project.  I
have posted the proposal to and posted the text
of the proposal below.


= Stratosphere =

== Abstract ==
Stratosphere is an open source system for parallel data analysis.
Stratosphere deeply integrates MapReduce and database technologies to
provide expressive and optimizable programming interfaces and at the same
time efficient and scalable execution.

== Proposal ==
Stratosphere is an open source system for expressive, declarative, fast,
and efficient data analysis. Stratosphere combines the scalability and
programming flexibility of distributed MapReduce-like platforms with the
efficiency, out-of-core execution, and query optimization capabilities
found in parallel databases.

== Background ==
There is currently a need for general-purpose cluster computing platforms
that are compatible with the Hadoop ecosystem, are more efficient, easier
to use, and can support more applications than Hadoop MapReduce, but are
not restricted to a specific data model and language (such as the
relational model and a variant of SQL). Stratosphere fulfils these needs.

Stratosphere exposes expressive APIs in Java and Scala (conceptually
similar to Spark, Cascading, Scalding) that allow arbitrary user-defined
functions in the same language and data model that the program is written
in. Stratosphere programs pass through a cost-based optimizer that finds
the best execution path for these programs depending on the data and
cluster characteristics. The design and implementation of Stratosphere is
based on research that generalizes query optimizers in relational
databases. Stratosphere has a distributed runtime that is architected upon
the principles of parallel databases, providing true pipelining (a basis
for stream processing) and efficient out-of-core algorithms for grouping,
sorting, joining, and aggregating data. Stratosphere provides first-class
support for iterative algorithms via a built-in iterate operator, covering
Machine Learning and graph analysis use cases. It achieves performance
similar to Apache Giraph without being a specialized graph processing

Stratosphere has undergone three major releases (v0.1, v0.2, v0.4) and some
minor ones.

== Rationale ==
Stratosphere started out in 2008 as a research project by the Technical
University of Berlin, the Humboldt University of Berlin, and the Hasso
Plattner Institute, and has received subsequent funding from the German
Research Council, the European Institute of Innovation and Technology, the
European Commision, and industry.

The traction of Stratosphere has by far exceeded our initial expectations,
and we are therefore seeking an organizational long-term home for
Stratosphere beyond the University walls that will house and further
encourage contributors from companies and other organizations that are
interested in Stratosphere. We believe that the Apache Software Foundation
is the ideal home for Stratosphere. Stratosphere integrates with several
existing Apache projects, such as HDFS, YARN, HBase, and Avro. The team is
familiar with the Apache processes and fully subscribes to the Apache
mission. One of the proposing members is a long-time Apache contributor and
PMC member.

== Initial Goals ==
 * Move the existing codebase to Apache
 * Integrate with the Apache development process
 * Ensure all dependencies are compliant with Apache License version 2.0
 * Incremental development and releases per Apache guidelines

== Current Status ==
=== Meritocracy ===
Stratosphere operated on meritocratic principles from the get go. The
initial project proposal submitted to the German Research Council
in 2008 stated that all code developed in the project will be released as
open source under the Apache 2 license. Currently, all the
discussions pertaining to Stratosphere development are public on [[|GitHub]]  and our [[!forum/stratosphere-dev|mailing list]].
The current incubation proposal includes the major code contributors to
Stratosphere. Several additional people have worked on the Stratosphere
codebase for research prototypes and industry use cases and would be
interested in becoming committers. We are starting with a small committer
group and we plan to add additional committers following an open
merit-based decision process during the incubation phase.

=== Community ===
Currently, the core of Stratosphere is developed at TU Berlin, mainly by
the committers listed in this proposal. Additional people from several
Universities and companies in Europe are working with Stratosphere and are
interested in becoming committers to the project.

During the years, Stratosphere has been adopted as a platform for research
and teaching in several Universities (TU Berlin, HU Berlin, HPI, RWTH,
Inria, KTH, U. Trento, UCSD, and others), and it is currently witnessing
its first industrial installations. We are seeing a rapidly growing
interest in Stratosphere by both startups and large companies, as well as a
growing community (our first [[|Stratosphere Summit]] in
November 2013 attracted over 80 participants). Stratosphere was recently
accepted as a mentoring organization in Google Summer of Code 2014.

We believe that acceptance in the Apache Software Foundation will
consolidate the current community under one organizational umbrella, and
most importantly accelerate the growth of the community.

=== Core developers ===
The core developers of the system are Stephan Ewen, Fabian Hueske, Daniel
Warneke, Robert Metzger, Ufuk Celebi, and Aljoscha Krettek, who are all
committers in the current proposal.

=== Alignment ===
Stratosphere is compatible with, and related to several Apache projects.
Stratosphere re-uses parts of Apache Hadoop, in particular HDFS and YARN,
as well as Apache HBase and Apache Avro. Stratosphere is a very good
compilation target for query languages such as Apache Hive and Apache Pig.

== Known Risks ==
=== Orphaned Products ===
There is strong interest in Stratosphere by several companies and
organizations, and there is currently a long-term commitment to fund
salaried developers for Stratosphere by public and private organizations in

=== Inexperience with Open Source ===
Sebastian Schelter is a committer and PMC member of Apache Mahout and
Apache Giraph, member of the Apache Software Foundation, member of the
Incubator PMC and project mentor for Apache Drill. Sebastian, along with
our mentors, will guide the rest of the committers that have experience
with releasing software as open source but little experience in
participating in an open source project besides Stratosphere itself.

In mid-2013 Stratosphere transitioned from an "open source project with
publicly accessible source code" to an open source project that puts the
community first. We moved from a University-hosted git repository to
GitHub, where we discuss all issues publicly. This also includes release
planning (via GitHub's milestone feature) and code reviews. We also moved
our build system to the publicly available Travis-CI. The mailing lists are
hosted with Google Groups, we use the public Maven repository
infrastructure of Sonatype. The source code of the
www.stratosphere.euwebsite is publicly available and is meant to be
changed by external
contributors (for example for documentation purposes).

=== Homogeneous Developers ===
Most committers in this proposal belong to the same institution (TU
Berlin). The engagement of these committers goes well beyond the necessary
development to support research, and all committers work on Stratosphere in
their free time. Several people from other institutions are working on and
are familiar with the Stratosphere codebase. We will work to attract them
as future committers during the incubation phase, following a merit-based

=== Reliance on Salaried Developers ===
Currently, Stratosphere receives support from salaried developers, in
particular from graduate students at TU Berlin that are funded by the
German Research Council, the European Institute of Technology, and the
European Commission. These students work in their free time on Stratosphere
in addition to their employment.

We expect that Stratosphere development will occur on both salaried and
volunteer time. We will recruit additional committers, including
non-salaried developers, and we will work to ensure that the project will
move forward independently of salaried developers.

=== Relationship with Other Apache Products ===
Stratosphere interfaces with several existing Apache projects: Apache HBase
for storage, Apache Hadoop (HDFS for storage, YARN for resource management,
and Stratosphere contains a generic wrapper for Hadoop MapReduce input
formats), and Apache Avro (for serialization). Stratosphere uses Apache
Maven and Apache Commons libraries internally. Stratosphere can be a great
compilation target for Apache Pig and Apache Hive, although such
functionality is not yet implemented.

Stratosphere is also related with several projects undergoing incubation in
the Apache Incubation project, such as Tez, Drill, and Spark (graduated).
While all these projects target sufficiently different spaces and have
different architectures, it would be interesting to explore code reuse
possibilities. For example, we are currently basing our design for
compiling SQL to Stratosphere on the Optiq library, also used by Apache

=== An Excessive Fascination with the Apache Brand ===
We believe that the Apache brand will help us attract contributors to
Stratosphere, by giving us a well-defined, transparent development process
under a known brand. At the same time, Stratosphere already has a healthy
community and current funding guarantees the further codebase development
and growth of the project for the next 3-5 years. The reason for this
proposal is not to gain publicity, but to further strengthen the longevity
of the project as explained in the Rationale section.

== Documentation ==
 * [[|Project website]]
 * [[|Documentation]]
 * [[|Codebase]]
 * [[!forum/stratosphere-dev|Mailing list]]

== Initial Source ==
Stratosphere is hosted on [[|GitHub]] . This is the
codebase that we will migrate to the Apache Foundation. The code was
previously hosted on a TU Berlin's own git infrastructure. It has always
been Apache 2.0 licensed.

=== Source and Intellectual Property Submission Plan ===
All initial and past committers will sign a CLA with the ASF while the
incubator proposal for Stratosphere is being discussed. All organizations
that have employed Stratosphere contributors in the past will sign a SGA.
Current contributors will sign a CCLA. All major contributors are still
active in the project.

=== External Dependencies ===
All critical dependencies are, to the extend of our knowledge, from other
Apache projects. These include Apache Hadoop (for YARN and HDFS) and some
libraries (log4j, commons codec, junit and more). Our web frontend uses
some MIT-licensed JavaScript libraries.

== Required Resources ==

=== Mailing list ===
We will migrate our mailing lists to the following:

=== Source control ===
We would like to use Git for source control and enable GitHib mirroring
functionality, where code reviews on GitHub are automatically
forwarded to the developer mailing list. (See also: [[]

=== Issue tracking ===
We are currently using GitHub for issue tracking. We request an
Apache-hosted JIRA, and we will import existing issues there.

== Initial committers ==
 * Stephan Ewen -
 * Fabian Hueske -
 * Daniel Warneke -
 * Robert Metzger -
 * Ufuk Celebi -
 * Aljoscha Krettek -
 * Kostas Tzoumas -
 * Sebastian Schelter  -

=== Affiliations ===
 * Stephan Ewen (TU Berlin)
 * Fabian Hueske (TU Berlin)
 * Daniel Warneke (Amadeus IT Group)
 * Robert Metzger (TU Berlin)
 * Ufuk Celebi (FU Berlin)
 * Aljoscha Krettek (TU Berlin)
 * Kostas Tzoumas (TU Berlin)
 * Sebastian Schelter (TU Berlin)

== Sponsors ==
=== Champion ===
Alan Gates (

=== Nominated Mentors ===
 * Sean Owen ( (Note: Sean is an Apache member but not
currently on the IPC, he will need to request IPMC membership)
 * Ted Dunning (
 * Owen O'Malley (

=== Sponsoring Entity ===
The Apache Incubator

NOTICE: This message is intended for the use of the individual or entity to
which it is addressed and may contain information that is confidential,
privileged and exempt from disclosure under applicable law. If the reader
of this message is not the intended recipient, you are hereby notified that
any printing, copying, dissemination, distribution, disclosure or
forwarding of this communication is strictly prohibited. If you have
received this communication in error, please contact the sender immediately
and delete it from your system. Thank You.

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

Sanjiva Weerawarana, Ph.D.
Founder, Chairman & CEO; WSO2, Inc.;
email:; office: (+1 650 745 4499 | +94  11 214 5345)
x5700; cell: +94 77 787 6880 | +1 408 466 5099; voip: +1 650 265 8311
blog:; twitter: @sanjiva
Lean . Enterprise . Middleware

View raw message