[postgis-users] Re: Hosting Co-op (was PostGIS hosting?)

[Frank Warmerdam]

Paul Ramsey wrote:
> As a system administrator, my take is that while things would work well 
> in the beginning, in the end things would get nasty with version 
> interdependencies and the requirements of different users. If I upgrade 
> the system-wide PHP installation and break it for some of the users... 
> well, they will not like me much. Similarly for systemwide installations 
> of GEOS, PostgreSQL, mapserver, etc. While some people will simply use 
> their area for happy tinkering, others will want to deploy stable 
> applications and leave them be. The tinkerers would be happy to always 
> have the latest version of everything, but the deployers would want 
> things left well enough alone.

Tyler / Paul,

I would tend to agree that for this to be practical versioning would be an
issue.  I think it might be best to support multiple versions as time
progresses with old versions not being retired untill there are no paying
co-op members using the version any more.

More generally, I like the idea of a co-op but I am dubious about whether it
would make sense.  Whether it would get enough supporters to justify the effort.
Or alternatively, if there might be some folks interested who want a higher
degree professionalism than a cooperative would be able to provide.

For instance, I think that network load, and disk space required would vary
quite radically for different users and that it would be relatively hard to
determine which users are using how much (of network bandwidth at least).
Coming up with a fee structure that reasonably reflects this would be
challenging.  In fact, one heavy user might easily heavily load the machine
or run up big extra bandwidth charges.

Caveats aside, I think it would be very nice to have a cooperatively managed
hosting system available.  From my point of view, I would like *one* of the
available software sets to closely track the current development version of
some of the key packages (PostGIS, MapServer, and GDAL/OGR for instance).

Hopefully the system would attract some supports as a convenient place to do
prototyping.   To test out MapServer and the various component technologies
even if the intent would be to eventually deploy their production systems
elsewhere once they are satisfied.  The nice thing about doing some prototyping
on a relatively public system like this, is that alot of the starting headaches
getting the environment setup would be skipped.  Also, it would be easier for
outside developers to diagnose issues on a such a semi-public system.

One question that comes to mind, is what sort of components you would want to
provide.  Ideally I would like to see:

  o PostgreSQL/PostGIS (with GEOS, PROJ, etc).
  o MapServer with WMS/WFS/WCS client and server, PROJ, pdf, flash, GDAL, OGR
  o Apache 2
  o Python MapScript bindings, possibly Zope?
  o Perl MapScript
  o PHP/Mapscript
  o the full suite of PHP MapScript based Chameleon/CWC2/MapLab technologies
  o GDAL/OGR with as many extras built-in as possible builtin, including
    JPEG2000, OGDI, HDF, GML (Xerces), DODS, and Python bindings.
  o MySQL (plus spatial extensions)

I think it would also be ideal to provide a full Java environment.  I
don't know this side of the world so well, but I assume that would include
"Apache Tomcat", Java MapScript, GeoServer, Java classes for interfacing
with PostGIS, other GeoTools2 related stuff.

I think setting up a formal cooperative with fees might be pretty organizationally
challenging.  It would be easier if someone - perhaps someone who already needs
to setup a co-hosted server - would be willing to operate the system as a
commercial effort even if not with the intent of earning alot of profit.  This
might be an independent consultant or perhaps a user group.  A company like
DM Solutions or Refractions would also be great, but it might kind of undermine
their more professional support options.   For a consultant, or consulting
company operating such a hosting platform could hopefully be a valuable source
of recognition/marketting even though the hosting solution itself isn't likely
to be be too lucrative.

If someone does decide to launch such an effort, I would be willing to put in
time and effort to help set up many of the software packages, primarily to
promote the various technologies, but also possibly for some free hosting and,
as always, for the recognition that can come from such an effort.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent