Background

Power Outage occurred quite frequent in Malaysia especially in year 2013. Refer to http://en.wikipedia.org/wiki/Power_outages_in_Malaysia, there were 4 times of blackout between May, June and July 2013.

Disaster recovery is different from Disaster avoidance. DA is about how to prevent a disaster (no downtime), and in the other side, DR is about how to recover whereby the system is already down.

A typical example of DA on virtualized environment is long distance vMotion between 2 nearby DataCenter. The problem of this approach is when the actual disaster strikes, there might be nothing to vMotion as Production is down.

Solution

DR solution has to be:
1. Simple. In fact, fully automated. No manual MS Word run book required.
2. Predictable. We know the fully automated DR will work when actual DR strikes because we do full test frequently (not quarterly as it’s too long).
3. Complete. It should fail over the entire data center, not just selected servers.

Solution: Isolating

Each of the the above VLAN will need their own VLAN too.
2 Distributed Virtual Switch required for this solution, and it’s quite costly approach

Pre-Failover

Post-Failover