This is a good case study from High Scalability on the operation of the microblogging site, Tumblr. The site is growing at 30% month over month and has about 500 million page views/day. This means storing around 1-3TB of data into Hadoop clusters daily!
As you can imagine, with this type of site, there were some growing pains. By all accounts, it is harder to scale than Twitter, yet the Tumblr equivalent of a Fail Whale isn’t an internet meme. 🙂 For us though, it is an informative look at their lessons learned while scaling.
Aside from the fascinating glimpse into the technical backend of the site, it provides some insight into the startup environment in NYC, and their hiring practices for new talent as well.
Snaps from around the office