You can read the previous article in the series here.
Disaster recovery (DR) is a fundamental best practice for all IT teams, yet many of them struggle with the technologies, tools, and processes they have.
The combination of data growth, the difficulty writing, updating and testing DR plans, and the need to make DR more cost-effective is making it very difficult for IT teams to do the job the way they know it needs to be done. Solutions like remote replication work well to reduce RPOs and RTOs for a limited number of mission-critical applications, but the expense of owning and operating dual environments for replication means that a lot of data does not get the DR coverage that the organization needs.
The Microsoft HCS solution is based on the hybrid management model where deduped fingerprints on a source CIS system are uploaded to Windows Azure Storage where they can be downloaded to another recovery CiS system for DR purposes. The recovery data that is stored in the cloud does not consume floor space, power or cooling costs in any of the organization’s data centers. Fingerprints in Windows Azure Storage are protected in the cloud by replication and geo-replication services. One of the key management elements is an object called the metadata map, which contains pointers to all the fingerprints that were uploaded by the CiS system. The combination of the fingerprints and the metadata map creates a portable, deduped data volume that can be downloaded to another CiS system during recovery operations.
In a recovery operation, the metadata map is downloaded first and then all the data that had been uploaded becomes visible to applications and users. Thereafter, the download process is driven by applications as they access their data. This deterministic, application-driven recovery process limits the data that is downloaded to only the deduped working set, leaving all the data that is not needed in the cloud. The thin, fast recovery capabilities of the Microsoft HCS solution enable IT teams to test their DR plans without disrupting their production operations. Recovery times with deterministic restores are short. Taking cloud snapshots several times a day can reduce recovery points.
The hybrid cloud management model enables a number of flexible, cost-reducing data recovery architectures. A single CiS system can be a spare for other CiS systems in a N:1 topology, or one or more CiS systems can be used to recover data for one or more disaster-stricken CiS systems in a N:N topology. There is no need to duplicate a data center environment for DR with the Microsoft HCS solution.
The flexibility and leverage gained through the hybrid cloud management model does not end with DR scenarios, but extends to other aspects of storage management as well. Our next bonus article, BONUS #3: Taming the capacity monster, continues the exploration by showing how the same fingerprints that were uploaded to Windows Azure Storage and used for DR purposes are also used to extend the capacity of on-premises CiS systems.