ABSTRACT
In case of server problem, when servers have storage directly attached and servers fail, that data is offline until that server is recovered. In the worst case, the data is recoveredfrom snapshot or from tape onto another server. The impact of server problem is that the data is unavailable until another server running the same application accesses the storage. In the case of drive failure, If the RAID is mirroring, it simply runs on the other copy of the mirrored data with no performance impact—but it’s like driving without a spare tire. A tiering approach will avoid every single load being assessed as “critical” and lead to a situation where the data center auxiliary generation system is overloaded and requiring an expensive improvement. Disaster recovery comes in different shapes and sizes: server failover, also known as multipathing; near continuous data protection, also known as asynchronous replication; and full continuous data protection (CDP), also known as synchronous replication.
