[OpenLayers-Users] what does a good FOSS WEB GIS app deploy look like?

[Paul McCullough]

All,

This topic also posted on geoserver nabble (different wording):
http://www.nabble.com/deploying-a-geoserver-stack-td15213615.html#a15213615

I am preparing to move my FOSS WEB GIS stack off my singular desktop box and
into an environment that will be more like what I envision for production. 
Before I decide what to put where I thought I'd ask what others are doing
and why.  The better the response, the more time I'll spend on this thread
and on preparing a summary.  I'll be up front - I am a mere mortal
developer, and not knowledgeable about networks or security.

I'll try to start the discussion by presenting my somewhat naive and
incomplete vision for my deploy.  To start off, any application
configuration is going to look differecnt depending on what the application
does and what are the constraints imposed by the adminstion group.

My Application
--------------
My application is a simple map browser.  It answers questions like: where is
the nearest polling station for this address. We anticipate lots of reads
against the tilecache. At this point, I do not know how we will be drawing
things that are not in the tile cache or things that may overwhelm
javascript. I expect that at some point we may be making WMS and WFS calls
directly to geoserver.

My Stack
--------
 - Apache/Tile Cache/mod_python 
 - Geoserver/Tomcat
 - webservices (php/java)
 - PostGIS

My Deploy
---------
I will be deploying to a single datacenter inside the firewall.
I am planning on using SUSE on VMWare.  
And I am planning on setting up a VM for each component from the stack.
So for example, given the list just above, I will have 4 VMs.
Why have a separate VM for each component?
The thought is that it should make patches and upgrades more independent.
In other words, if PostgresQL x.x needs SUSE x.x, I can accomodate that
without pertubing any other component.
I am anticipating that this will help split out the admin responsibilites
more cleanly.
Is this worth the added complexity?
I don't know?  Any comments from people with experience?

Useful Links (not all of which apply to my deploy)
--------------------------------------------------
geoserver on ubuntu
http://grimmeister.wordpress.com/2007/08/08/setting-up-an-open-geospatial-consortium-service-server/

deploying geoserver in production
http://docs.codehaus.org/display/GEOSDOC/6+GeoServer+in+Production+Environment

Comment from Chris Holmes of TOPP here
http://www.nabble.com/deploying-a-geoserver-stack-td15213615.html#a15213615
 
geoserver outside the firewall
https://www.seegrid.csiro.au/twiki/bin/view/Infosrvices/GeoserverArchitecture

using javascript to retrieve information from remote domains via an
XMLHttpRequest
http://trac.openlayers.org/wiki/FrequentlyAskedQuestions#ProxyHost

Any and all comments are welcome.
Thanks,
Paul

-- 
View this message in context: http://www.nabble.com/what-does-a-good-FOSS-WEB-GIS-app-deploy-look-like--tp15238589p15238589.html
Sent from the OpenLayers Users mailing list archive at Nabble.com.