If the search feature is slowing down the site, I think using Google would be a satisfactory permanent alternative. We've been using it for years at the PR Watch web site, with no complaints.
I'm very much against this. Having search results in realtime is of crucial importance for serious article work to avoid duplicates. The Google spider only comes ever so often and misses a lot. PR Watch is one thing, Wikipedia is a couple orders of magnitude bigger, though. As a very, very temporary hack, the Google search is fine. But we need to get the realtime search (with wikisource no less!) back up and running ASAP.
If we have to disable critical features to keep the site running, we may as well disable editing altogether until we have raised enough money to buy a bigger server and/or hire developers.
As for optimization, there are still a few things we need to do, but we also have to keep in mind that a single server is now running all language Wikipedias + meta, and all of them are growing fast (faster now that Google indexes us again). I hope we have the nonprofit setup soon, and if we can't get it setup, we should start raising donations without an organization. Kuro5hin raised 35000 dollars in a few days -- I'm not sure if we can get that high, but we should certainly try.
Regards,
Erik