Given the increased threats facing organizations today, a single DR site is no longer good enough. Instead organizations need to apply a hybrid cloud mentality to their disaster recovery strategies.
What Will You Fail?
The most important data after a disaster is the data that was in use at the time just prior to the disaster. For most organizations that means that less than 15 percent of the total data set needs to be available in the disaster recovery site. Of course that 15 percent was active for a reason, most of it is critical and that means that most of it needs quick recovery.
If we can eliminate the DR data set to 15 percent of total data, that likely means the organization can afford to use an advanced data protection technique like replication to move data as it changes to the DR site. The result is our DR site is typically only a few minutes behind the live data set.
Where Will You Fail?
Hybrid DR is the ability to recover an application in a variety of locations based on the threat. Despite popular belief the primary recovery location should be on-premises. Most “disasters” are not building evacuating events. Most outages are the result of a software bug or hardware failure, most common is an application corruption or storage system failure. Why fail to the cloud or another location if most of the data center is still in-tact?
Data centers should replicate their most active from on-premises primary storage (15 percent of total data) to a single secondary repository also on-premises. That system should be able to provide adequate performance so it could “stand-in” if a primary storage system goes offline. This system covers 99.9 percent of failures. Most organizations don’t face a data center wipe out more than once a decade.
Even though there is only a 0.01 percent chance of failure, the consequences of a complete data center loss does require planning. Here, data centers should look to replicate data from their on-premises “stand-in” storage system to a another location.
There are potentially two type of remote locations to consider; close and far.
Some organizations need a very low recovery point objective (RPO) and recovery time objective (RTO). They’ll want a “close” location, so that network latency is minimal. They’ll still want a “far” location too, in case of a massive regional disaster.
These remote locations need to not only have a very recent copy of the data, but also available compute to host the organization’s applications. This dual requirement (data and compute) make service providers and the cloud ideal choices. An organization may decide to use both. A regional service provider for “close” sites and a public cloud provider for the “far” site.
The need to perform disaster recovery on-premises, at a service provider and in the cloud is the very definition of hybrid cloud DR. The combination provides the organization with the lowest possible RPO/RTO while providing the broadest level of protection against a variety of disasters.
To learn more about replication and the hybrid cloud check out our video, “The Next Best Steps for Hybrid Cloud”.