PostGreSQL load increasing over time, why?

It's a CentOS server (I don't know the specs) and just before anybody states the obvious, keep in mind these mitigating factors:

• the server does a nightly VACUUM job
• all the tables are indexed
• it's pretty much read only (meaning the DBs are not increasing in size)
• the number of queries being ran has been the same every month

Here's a graph of the server load:alt text http://img96.imageshack.us/img96/6352/loady.gif

UPDATE (2010/9/15): We never found a server or config cause to this, but our application was optimized and the load is back to acceptable levels at the same usage:

1. Cached some data
2. Foun开发者_开发技巧d some queries that were rewritten
3. Found some redundant queries

Not only should you vacuum, but also vacuum analyze, and reindex (at least once a month), since indexes are not vacuumed.

May I also suggest that your piece of software is just so good that you gain more users over time?