When a primary cluster falls in OpenSearch, the data recovery process is usually initiated, but it might take time to complete. This means that you’ll suffer from downtime in the system and data will not be available to you users. This is especially problematic if you’re running a mission-critical deployment and if your SLA dictates constant availability.
It’s almost impossible to avoid downtime like this if you’re working with OpenSearch on your own. Theoretically, your application would need to automatically write to multiple clusters and support partial errors and other edge cases that might occur while mirroring.
Opster’s Multi-Cluster Load Balancer provides active-active, zero downtime data recovery. The MCLB mirrors traffic routed to the cluster to designated backends in real time, providing two identical copies of the same index on two different clusters, completely simultaneously. There’s no time delay, and therefore no data loss, no matter when the disaster occurs.
When one data center is down, you won’t experience downtime because the other center will still be up and running with the exact same data, available and serving searches to you and your users. The Load Balancer mirrors all requests in an active-active manner in order to allow this.
If the primary OpenSearch cluster is not available, the MCLB will route indexing and search operations automatically to the follower ES cluster, providing a real time active-active experience with no downtime.