Really interesting post about the architecture at Tumblr, which has changed a lot over the past few years, and is a fascinating selection of tools stacked together. Especially good on the reasoning behind tool selection.

"My dream cloud interface is not about booting virtual machines and monitoring jobs, but about spending money so my job finishes quicker. The cloud should let me launch some code, and get it chugging along in the background. Then later, I would like to spend a certain amount of money, and let reverse auction magic decide how much more CPU & RAM that money buys. This should feel like bidding for AdWords on Google. So where I might use the Unix command “nice” to prioritize a job, I could call “expensiveNice” on a PID to get that job more CPU or RAM. Virtual machines are hip this week, but applications & jobs are still the more natural way to think about computing tasks." Yes, this. And: lots of people _think_ the cloud works like this, but it really doesn't, yet. Parallelization/adding computing power is more practical, but it's not been made easy like a bunch of other things have (so far).

"We were a niche site and in the course of eighteen months had siphoned off about six thousand users from our massive competitor, a pace I was was very happy with and hoped to sustain through 2011. But now the Senior Vice President for Bad Decisions at Yahoo had decided to give us a little help." Maciej on what Scaling Pinboard Fast actually looked like. Some good anecdotes in here.

Some well-worn tales here, but also some good new ones, particularly when it comes to query-profiling and all forms of caching.

A few notes on Flickr's queueing systems.

I am amazed that anyone would have the patience to respond to Mike Arrington's general arsery, but it seem the Twitter team do. They are better men and women than I.

Marc Andressen provides some interesting analysis of F8. The news for popular applications, such as iLike, isn't good.

"Facebook has roughly 200 dedicated memcached servers in its production environment". Blimey. 200 x 16GB is a LOT of cache. Still, 99% of hits go straight to cache - impressive!

Simon is right: nice to have all this lot in one plsace.

Mmn. Fragment caching. I knew I didn't want page/controller caching, was dreading this... and it all looks really rather doable. Sorted!

Nice write-up of Cal's tutorial