Thoughts on Digg and Cassandra
It's no secret that the latest version of Digg's site, Digg v4, has had some serious hiccups since its launch. So many, in fact, they've created a page specifically for their Known Issues. It's been suggested that most of their technical problems have been in part due to switching their database from MySQL to Cassandra.
Now I tend to have grace for early adopters. I'm an addict for new technology, myself. I'll see a benchmarks or a feature set of a "promising" technology and I'm sold. This, of course, is a luxury I can afford given the fact that I do not operate a site that generates 30 million unique visitors a month and has $40 million in venture capital. But It wasn't a luxury Digg could afford.
Hopefully this will serve as a lesson: If the shoe doesn't fit, find a different pair. Cassandra is used by Facebook (it's creator), Twitter, Reddit and more, but not in the same capacity as Digg. Most of Cassandra's use has come in the form of caching and indexing. Digg is using it as a replacement for their MySQL database. This, as it turns out, is not an ideal solution (or at least its implementation is not).
I guess this is the part where I tell you that Digg should have used MongoDB... So... Digg should have used MongoDB.