So I just got in to San Diego yesterday, and i’ll be here til Friday. I’ve brought along the ol’ Powerbook so I can get some Freetag hacking on in the evenings. I’m currently attending Cal’s Flickr scaling presentation, which is pretty detailed, and the audience seems to be pretty engaged.
Interesting points so far:
- A nice, quick overview of available software load balancers.
- Not much of a discussion so far of the dynamic structure of Flickr’s API, but some hints that the approach lies between monolithic functions and full OOP (a difficult balance to achieve).
- Short discussion of usefulness of Layer 7 caching. Separation of hashed content in caches and ACLs would probably allieviate some of the performance issues that we regularly see on Upcoming. Luckily, we are at the point where we are starting to optimize some of our process, and so far, the one major point i’d make is to avoid left joins whenever possible.
One nagging question I have: Where can I sample a trifle in the US, Northern California, specifically?