Performance: Fast Recovery

Allows to restore full cluster operation and high availability quickly after a component failure is resolved.

Intro 

Modern world business success directly depends on its applications and services availability. These should run 24/7/365 and the storage backend should be able to support this by providing High Availability. This means not just being able to withstand a component failure but also to resume full operation as soon as possible. 

Problem 

Synchronous replication ensures that data is available at multiple nodes at the same time. It increases redundancy of the storage subsystem and, as a result, the uptime for applications and services. Nonetheless, replicated storage can get out of sync because of hardware or software failures while synchronizing large amounts of data may take a long time. 

During the synchronization process, applications are prone to downtime since storage is only available on synchronized nodes. Subsequent hardware or software failures during this time can cause storage to go offline. 

Also, performance of the storage subsystem is lower because of additional workload, which is introduced by the synchronization process. 

Solution 

In order increase applications and services uptime, even during the synchronization process, and minimize the risks of downtime, StarWind VSAN allows storing synchronization journals on separate persistent storage. In this scenario, StarWind will only synchronize changes written to the storage while one of the partners was out of sync, which minimizes the amount of time and data needed for synchronization. 

Both sides will become available within minutes, which increases the uptime of business-critical applications. 

Moreover, performance impact is minimal. StarWind partner nodes do not need to synchronize data which was already written on every node in the cluster. 

Conclusion 

To maximize business applications and services uptime, StarWind Virtual SAN ensures fast recovery of replication partners. This allows not only withstanding node failures but also resuming storage High Availability in a very small amount of time once partner nodes are online.