About data store synchronization
Deployments requiring the highest levels of redundancy and performance should employ data store synchronization between designated master and backup appliances. Data store synchronization enables pairs of SteelHeads to synchronize their data stores with each other, even while they’re optimizing connections. This feature is typically used to ensure that if an appliance fails, no loss of potential bandwidth savings occurs, because the data segments and references are replicated on the backup SteelHead.
You can use this feature for physical in-path, virtual in-path, or out-of-path deployments. You enable synchronization on two SteelHeads, one as the synchronization master, and the other as the synchronization backup. Traffic for data store synchronization is transferred through either the primary or auxiliary network interfaces, not the in-path interfaces.
Data store synchronization is a bidirectional operation, regardless of which deployment model you use. The SteelHead master and backup designation is only relevant in the initial configuration, when the master SteelHead’s data store essentially overwrites the backup appliance’s data store. The synchronization master and its backup:
• must have the same hardware or virtual model.
• must be running the same software version.
• don’t have to be in the same physical location. If they’re in different physical locations, they must be connected with a fast, reliable LAN connection with minimal latency.
When you have configured the master and backup appliances, you must restart the optimization service on the backup appliance. The master restarts automatically. The data stores are actively kept synchronized.
If one of the synchronized SteelHeads is under high load, some data might not be copied.
If data store synchronization is interrupted for any reason (such as a network interruption or if one of the appliances is taken out of service), the SteelHeads continue other operations without disruption. When the interruption is resolved, synchronization resumes without risk of data corruption.