Disaster Recovery Plan

For cases when imminent storms, or any sort of natural disaster is expected to occur, we temporarily provision redundancy for the instances located in the region’s location.

Case study: Superstorm sandy

1. We mass provisioned servers in Amazon US-West Oregon.
2. We setup replicas for each server hosted in US-East Virginia.
3. We let the replicas run until the storm had subsided, 3 days later.

For these cases, the user can opt to do one of two things:

1. They can opt to use the redundant replica, promote that to master, and make their current instance a replica.
2. They can wait it out, and act if something does happen.