I have been posting about the general slowness since I started as a customer 2 weeks ago. This included three different servers. At times, I could tell that it was not just processor-bound slowdown, based on the amount of overall cpu usage on the servers.
I decided to use one of my callbacks. I talked to a very knowledgeable person today. There was one issue that was short term, and one that represents the longer term problem.
Short term: cienega, and other servers, experienced incredibly slow throughput over the past day due to a failed network card which caused problems on a switch. This has now been fixed.
Long term: The problem that I have been seeing is related to their DNS servers. Dreamhost uses BIND, which is unable to handle 200,000 domains with acceptable performance. They have upgraded to 64 bit servers in order to increase addressable memory, but this only caused minor improvements. They are moving to a custom rdns solution, which may not be in place for a little while yet.
How has this affected me? I am using php with non pconnect mysql connections. (which is what they recommend, and also happens to be in use by wordpress and gallery) So every mysql server connection is using the horribly slow DNS lookup, and not caching any of the dns results.
A few posts have pinpointed DNS, but not with these specifics. Unfortunately, it is not an overnight solution. And those that said “too much growth” were correct, but not due to the most frequently cited causes.
I am in a position where I can wait this out and see if it does get any better. If I was hosting accounts for other people, or doing something commercial, I would have a tougher decision.