We’re on Tumblr. If you’re reading this you’re (probably) on Tumblr.
And if you’re on Tumblr and we’re on Tumblr we have a shared history of great times, fun people and… unfortunate downtimes.
These don’t happen as much as they used to in our neck of the woods.
And that’s because Tumblr is scaling… everywhere.
As in, it gets 15 billion page views a month, has a peak rate of 40 thousand requests per second, collects more or less three terabytes of new content a day, all running on approximately one thousand servers.
And they’re doing this with about 20 engineers.
If you’re a geek, a friend of a geek, or simply sympathetic (and/or empathetic) to geeks that build platforms and keep them running so that the rest of us can do what we do, read High Scalability’s article on Tumblr’s architecture, its goals, and the hurdles it’s facing as it tries to reach those goals.
In it, Blake Matheny, Tumblr’s Distributed Systems Engineer, guides us through stats, software, hardware, architecture and lessons learned.
It’s mesmerizing, in a geeky sort of way.
In the meantime, it’s Valentines tomorrow. Consider sending the Tumblr crew a slew of hugs and kisses.