Yep, this was discussed over on that thread too.
I'm over here in "Comparisons and Advocacy", using more of a style
of discourse that speaks to CIO and upper management.
I hear the BIG story a lot these days that many are buying into,
in terms of OTHER products like WebCT, Blackboard and so on.
I could recite their sales pitches, which harp on this as a "cost
saving feature" of their products. I also recall hearing this from
someone from Educause who gave a speech maybe 3 years ago
about how "you had to run a big service and small colleges may
suffer". Well, I'm here to say it ain't so. You can either run small
or run large, the costs will scale more nearly linearly with the size
of the organization (this is also an argument for smaller colleges
running their own servers, no harder than running email).
What's a bit different in my story from a cluster is that I don't force
the codebase to be the SAME for everyone. I think that's a better
level of service, plus is much less contentious. I see more time
than you might think spent on determining both when to upgrade
and to what version here at UW.
I'm also trying to advocate for NOT changing Moodle to add the
features in to allow it to scale with lots of organizational levels.
That would add unneeded complexity, making it harder to implement
for smaller installations, not to mention more db tables and joins.
If it's already in there, fine, just don't add any more