Disaster recovery for OpenSearch
Learn about disaster recovery (DR) strategies for self-managed OpenSearch services in watsonx.data.
Disaster recovery (DR) ensures that OpenSearch data and indexes can be restored after failures such as node crashes, data corruption, or regional outages. In watsonx.data, DR for self-managed OpenSearch services is implemented using OpenSearch-native mechanisms.
The following DR approaches are supported:
Snapshot and restore
Snapshots provide point-in-time backups of OpenSearch indexes that are stored in external object storage such as IBM Cloud Object Storage or S3-compatible storage.
Snapshots are typically used for periodic backups and full-cluster recovery scenarios.
Cross-cluster replication
Cross-cluster replication (CCR) enables near real-time replication of indexes from a primary cluster to a secondary cluster.
CCR is recommended for workloads that require low recovery time objectives (RTO).
Disaster recovery considerations
- Define recovery time objective (RTO) and recovery point objective (RPO)
- Use geographically separate storage or clusters
- Secure backup repositories with proper access control
- Test restore and failover procedures regularly