ClusterLion – Automatic Switchover for NetApp MetroCluster
The use of MetroCluster technology is a common method of protecting data from failures. It uses synchronous mirroring to protect the data in a cluster and activates services on the survivor site to serve the mirrored data originally owned by the site affected by disaster.
A potential remedy is to operate a service at a third data centre which monitors both locations and in case of failure initiates service transfer to the remaining cluster partner. However this leads to increased costs and complexity, and still doesn’t fully eliminate split-brain syndrome.
If Always-On availability is needed, switchover needs to be automatic. This leads to the following challenges:
- Automatic transfer cannot take place in the event of a power failure.
- Storage clusters are unable to differentiate between an overall failure of the cluster partner or an interruption of the interconnects. Because there is doubt, no automatic switchover takes place.
- Split–brain complications exist when more than one cluster believes the others are dead and continues to access data. This can lead to data corruption.