How to Cope With Cloud CrashesMay 30, 2011
The folks at LearnBoost, a free classroom management suite of online tools, have blogged last month about how they use MongoDB among other things to help them replicate in the cloud to prevent exactly the same sorts of outages that hit Quora and others when Amazon’s Web Services went offline.
Guillermo Rauch writes that LearnBoost didn’t suffer any downtime, largely because they took some care to build in redundancy to their apps and cleverly
replicated their data across different Amazon regions and data centers…
"Instances can communicate with others in other availability zones in the same data center in a very fast and performant [sic] way," he writes.
They used various add-ons to MongoDB, including bash scripts to automate Redis key value stores and Node.JS Package Manager to manage all code updates. They thought carefully about redundancy from the beginning of their design.
It is worth taking a look at their framework to understand what they did, and perhaps implement something similar in your own cloud-based operations.