Our customer change.org hit the Digg home page today!
We took a few light punches and then stood up and handled the load gracefully enough that nobody on Digg commented on site performance. If you read Digg, you know that’s an accomplishment!
I’m really proud of the way our team stepped up and took care of this for the customer.
Jayson immediately identified the problem as not CPU overload, or DB overload, which are common performance constraints; instead, a memory leak in the app itself. Memory leaks happen to the best of customers, and we’re happy to help find them.
What’s really cool is that with our slice technology, Ezra was able to address the problem by restarting the application via cron events staggered across their slices. It’s all done gracefully and the load balancer routes around the slice when it’s being restarted, so from the outside, it’s just smooth and continuous operation.
In the end, it’s a great example of what we’re trying to accomplish here. We have cool technology, to be sure, but in the end it’s the people that count!