Request for comments: Infrastructure migration assessment

Jo_Walsh · September 12, 2006, 2:05pm

dear Incubators,

Attached is an document that I helped Tyler produce, assessing OSGeo
projects' infrastructure needs and what components will need to move
in order to effect migration off the CollabNet platform into an
environment which OSGeo can self-manage.

What is at question here:
* What components projects need - are there pieces missing?
* Are projects interested in buying into a "complete picture" or
simply want some OSGeo-managed components (LDAP, DNS, or CMS but not
code repository management?)

The biggest constraint is the MapGuide project infrastructure; when
the CN contract ends on Jan 1st, MapGuide needs to have something in
place that will provide at least the same services (SVN, mailing
lists, web presence) with an SLA.

What needs to be figured out by the end of the month:
* Setup costs (this is probably too important to leave to volunteer
energy)
* Maintenance costs (at least for critical services there needs to be
a support SLA)
* Timescale - what the critical path is.

I know that a lot of people are insanely busy with F0SS4G this week.
It's important to get clarity on what projects' concerns and
requirements are ASAP, though. Even work on putting this plan together
and working up a prototype web UI has been stalling because neither
Tyler or I have had the time to commit; it's getting to the point
where this can't stall for much longer.

I pasted the commentary Markus offered on the board list into
http://wiki.osgeo.org/index.php/Talk:Project_Infrastructure_Migration
and would suggest this is a good place to offer more feedback.

The alternative to pursuing this plan is
* projects continue to run and maintain their own infrastructure
* the foundation still needs to support CMS, mail, wiki for
collaboration and education/publishing needs and this needs an SLA
* MapGuide takes out its own contract to rebuild its infrastructure

And i wonder how PSCs and committee chairs would feel about that.

There's been so much talk about this; reserve over whether it's the
foundation's place to start reimplementing CollabNet. I don't really
see the instantiation of trac+svn, mailing lists and a simple drupal
or plone portal as being that big a deal, and that's already more than
the CN platform provides.

cheers,

jo

transition_plan_v2.pdf (150 KB)

neteler · September 12, 2006, 4:42pm

Dear all,

[repost from board list]
thanks for the nice proposal! Here some suggestions/comments:

- Mapbender loss of archives: this is absolutely dangerous and
   should be avoided at least for other projects (copyright infringement
   issues, authorships etc are often commented therein)
- telascience.org: you may add "geodata processing" to the list of
   usages
- Drupal:
  - do we know anything about its scalability (many users)?
  - how to mirror it to other web sites (GRASS has 25+ mirrors worldwide)
- Hosting questions: do we have cost scenarios for the hired staff
   (maybe I should know)
- automated smoke tests on telascience.org: do we have different
   operating systems there?

Infrastructure: please add
- telephone conference server (Asterix or YATE,..) to render communication
   free and improve sound quality
- a translation portal for software user messages (web based, e.g.
   rosetta/launchpad (still non-free, let's free it), KDE kartouche, ...)
- IRC CGI interface for all foundation projects to enable people behind
   firewalls to enter IRC (limit access to OSGeo related channels)

Best,
Markus

On 9/12/06, Jo Walsh <jo@frot.org> wrote:

dear Incubators,

Attached is an document that I helped Tyler produce, assessing OSGeo
projects' infrastructure needs and what components will need to move
in order to effect migration off the CollabNet platform into an
environment which OSGeo can self-manage.

What is at question here:
* What components projects need - are there pieces missing?
* Are projects interested in buying into a "complete picture" or
  simply want some OSGeo-managed components (LDAP, DNS, or CMS but not
  code repository management?)

The biggest constraint is the MapGuide project infrastructure; when
the CN contract ends on Jan 1st, MapGuide needs to have something in
place that will provide at least the same services (SVN, mailing
lists, web presence) with an SLA.

What needs to be figured out by the end of the month:
* Setup costs (this is probably too important to leave to volunteer
  energy)
* Maintenance costs (at least for critical services there needs to be
  a support SLA)
* Timescale - what the critical path is.

I know that a lot of people are insanely busy with F0SS4G this week.
It's important to get clarity on what projects' concerns and
requirements are ASAP, though. Even work on putting this plan together
and working up a prototype web UI has been stalling because neither
Tyler or I have had the time to commit; it's getting to the point
where this can't stall for much longer.

I pasted the commentary Markus offered on the board list into
http://wiki.osgeo.org/index.php/Talk:Project_Infrastructure_Migration
and would suggest this is a good place to offer more feedback.

The alternative to pursuing this plan is
* projects continue to run and maintain their own infrastructure
* the foundation still needs to support CMS, mail, wiki for
  collaboration and education/publishing needs and this needs an SLA
* MapGuide takes out its own contract to rebuild its infrastructure

And i wonder how PSCs and committee chairs would feel about that.

There's been so much talk about this; reserve over whether it's the
foundation's place to start reimplementing CollabNet. I don't really
see the instantiation of trac+svn, mailing lists and a simple drupal
or plone portal as being that big a deal, and that's already more than
the CN platform provides.

cheers,

jo