[ULS-SIG] Proposal: ULS-SIG

Duncan McGreggor duncan.mcgreggor at gmail.com
Sat Mar 14 23:21:02 CET 2009

Below is a paste of the original proposal for this SIG. The original is
available in the Meta-SIG archives here:

I am including it as the first email to the list as a matter of
reference for those browsing the archives :-)


To the Meta-SIG members:

Steve Holden, Jim Baker and I have been discussing an interest in and
need for a Python ULS-SIG mail list at Python.org. In this email, we
propose the creation of this list, provide some background information
on the topic, and include members of the Python community who have
expressed interest in said list, members we would consider to be
"founding members" of this SIG.


We would like to establish an official Python ULS-SIG. This would
accomplish the following:
 1) further Python's (and Jython's) place in the world of SOAs
(service oriented architectures) as a first approximation of ultra
large-scale systems;
 2) lend Python some credibility in the research and prototype phases
of ULS work;
 3) send signals to the ULS community of academic, business, and
government organizations that Python and its community of individual
developers, of software projects, and of businesses are all valuable
resources for ULS work and related research;
 4) provide a place for like- and future-minded Python
developers/researchers to engage in pertinent conversations concerning
issues of Python's use in systems of massive scale including the
research, development, prototyping, and implementations there of.

Per the instructions at
http://www.python.org/community/sigs/guidelines/, we submit the

Short Description: Ultra Large-Scale Systems

Long Description: This SIG exists in order to discuss the emerging
field of computing in the context of ultra large-scale systems and how
such systems relate to or can be built with Python.

With the exception of its use at Google, Python as a language gets
very little press in large systems deployments. Java saw the
innovations of such concepts as the Enterprise Service Bus and has
dominated the field of SOA
development for large enterprise deployments. However, thanks to the
work of the Jython team, this is not something that we need to compete
with nor fight in anyway, but rather with which we can join forces.
This is important due to the fact that, in many regards, success in
developing and deploying SOAs is a first step towards creating ULS
systems. Nevertheless, in order to make ultra large-scale systems a
reality, a great deal of research and prototyping needs to take place
over the next 10-20 years. We will provide a means by which members of
the Python community can keep in touch with progress towards making
ULS a reality and find focus for efforts which converge with their own
interests, including (among many others) multi-language integration,
leveraging existing community efforts, and simplifying or building
upon successful architectures.

Background Information:

The term "Ultra Large-Scale Systems" comes out of the research that
was done in 2005-2006 by the Carnegie Mellon Software Engineering
Institute. It entails the "creation" of systems of the next 10-20
years, systems that will be too large for current design, development,
management, and interaction practices. These systems will come into
being through amalgamated services united to provide solutions for a
particular (potentially very large) problem space. The official report
is rather long but is available for free
(http://www.sei.cmu.edu/uls/files/ULS_Book2006.pdf). Shorter summaries
are also available. There is  one that gives a fairly concise summary,
especially on slide 22 and 23
(http://www.sei.cmu.edu/uls/files/OOPSLA06.pdf). Though that
particular presentation's content is focused on defense applications,
these problems are of deep interest/concern to industries such as the
medical field and other organizations that will have extensive and
diverse infrastructure and services operating in diverse and
potentially hostile environments. Another presentation outlines the
research that needs to be actively pursued in order to make ULS
systems a reality (see pages 69, 38, 37 -- note that the dotted 6.x.y
number refer to sections in the official ULS report;

Interested Parties:

On their own behalf or on the part of their development teams, their
communities, or their employers, the following individuals have
expressed an interest in a Python ULS-SIG:

Steve Holden, Holden Web
Jim Baker, Zyasoft
Duncan McGreggor, ULS Evangelist
Gustavo Niemeyer, Canonical
Chris Armstrong, Canonical
Ralph Meijer, XMPP Evangelist
Christian Tismer, Stackless
Kristján Valur Jónsson, CCP (EVE Online)
Axel Angeli, SOA Evangelist/Conference Coordinator

Thanks for your time and we look forward to hearing your comments.

Duncan McGreggor

More information about the ULS-SIG mailing list