Forum OpenACS Q&A: Largest OpenACS community?

Collapse
Posted by James Thornton on
What's the largest OpenACS community, and what hardware is it running on?
Collapse
Posted by Jonathan Ellis on
How do you wish to define "largest?"
Collapse
Posted by James Thornton on
Most active users.
Collapse
Posted by Denis Roy on
Not sure if it's the largest but we develop and maintain a website for AIESEC which is an international student organization with members in more than 80 countries.

Some facts about http://www.aiesec.net

  • OACS 4.6.3 and .LRN 1.0
  • >120,000 registered users, about 50,000 are active
  • more than 20,000 users logged in within the last 24h
  • so far easily up to around 100 concurrent users

  • 1 load balancer
  • 2 web server (Single 2,8 GHz P4-Xeon MT, 2 GB RAM)
  • 1 database server (Dual-Opteron 242, 1,6 GHz, 4 GB RAM)

Until very recently, we only had one webserver (Dual-P3 933 MHz, 2 GB RAM) which would support up to 60 concurrent users but after that become quite unstable. The portal system of .LRN 1.0 is very resource-intensive.

Collapse
Posted by C. R. Oldham on
We have over 30000 registered users on our site.  Of those, approximately 10000 I would define as "active".  The webserver is a dual 800Mhz Xeon with 1 GB of RAM.  The backend is Oracle on a dual 2.6GHz Xeon with 4 GB of RAM.

We wish it was faster. 😊  Our testing indicates that most of the bottleneck is in the database.  We've done a lot of query tweaking for our internal packages, and that has helped.  I'm afraid we are now at the point where the only way to really make the site smoke is to drop serious cash on a bigger Oracle box.  We've also been thinking about moving to a 9i RAC configuration with two servers, but I'm unsure if that will really help us in the performance department, and that means *really* serious cash because you need some sort of SAN configuration, none of the infrastructure for which do we have in place...

Collapse
Posted by Denis Roy on
I think upgrading your memory would help. We had a similar problem some months ago and our first step was to upgrade RAM from 1 to 2 GB and make heavier use of util_memoize. It helped a bit, the webserver would support about 60-70 concurrent users instead of only 40-50 before getting awfully slow and unstable.