Serendeputy 2 Update, July edition
Happy July, everyone!
I’m making progress, but I’ve been mostly heads-down with client work the last few weeks.
The biggest change is that I’m moving from database-driven querying to search-engine-driven querying. I set up the alpha version using Postgres and hstore, and ran some tests to see where it would fall over. It turns out that it falls over at smaller load levels than I’d like.
I’m glad I ran that set of experiments when it was just me and a couple of folks playing with it, instead of in front of the whole world.
So, I’m implementing a version using Amazon CloudSearch for some of the heavy indexing lifting. Depending on how things work out, I should be able to see if this will carry me through by the middle of August. Then, I can fire everything up again and see where the new constraint in the system will live.
It’s also giving me a good chance to go back and refactor a few of the pieces that I didn’t get right the first time. I need to tear everything down anyway, I might as well fix the other stuff while I have the tools out.
I’m hoping for beta one at some point between now and eventually. It’s not quite turning into Project Xanadu, but I worry that we’re trending that way.