High reliability
Alibaba Cloud Elasticsearch protects your data and keeps clusters available through two mechanisms: snapshots for backup and disaster recovery, and load balancing for distributing traffic across data nodes.
Use these features to:
-
Back up cluster data automatically on a daily schedule
-
Save snapshots to Object Storage Service (OSS) for long-term retention and cross-cluster recovery
-
Restore data from a snapshot to recover from accidental deletion or hardware failure
-
Distribute client requests evenly across all data nodes to avoid overloading a single node
Snapshots
Elasticsearch snapshots are point-in-time backups of your cluster data. Alibaba Cloud Elasticsearch supports two snapshot types:
| Automatic snapshots | Manual snapshots | |
|---|---|---|
| Storage location | Alibaba Cloud Elasticsearch (managed) | Your OSS bucket |
| Retention | Last 3 days | Controlled by you |
| Restore target | Same cluster only | Any cluster in the same region as the OSS bucket |
| Use when | You need daily automated backups with simple restore | You need long-term retention, cross-cluster migration, or fine-grained control |
Automatic snapshots
Alibaba Cloud Elasticsearch automatically creates daily snapshots. Enable this feature on the Snapshots page in the Elasticsearch console, and set the time that works best for your workload.
Alibaba Cloud Elasticsearch retains only the snapshots from the last three days. Automatic snapshots can only be restored to the cluster where they were created.
To enable automatic snapshots and restore data from them, see Data backup overview and Create automatic snapshots and restore data from automatic snapshots.
Manual snapshots
Manual snapshots are saved from your Elasticsearch cluster to an OSS bucket. Use manual snapshots when you need retention beyond three days, or when you need to restore data to a different cluster.
Before creating manual snapshots, activate OSS and create a bucket in the same region as your Elasticsearch cluster.
You can store snapshots only in OSS buckets of the Standard storage class. The Archive storage class is not supported for snapshot storage.
Create snapshots for specific indexes by calling the snapshot API. To restore index data, call the restore API to load data from any snapshot in the OSS bucket.
By default, each data node processes up to 40 MiB of data during restoration. To adjust this limit, modify the max_restore_bytes_per_sec parameter. Manual snapshots can only be restored to a cluster in the same region as the OSS bucket. For details, see Snapshot and restore in the open-source Elasticsearch documentation.
For step-by-step instructions, see Create manual snapshots and restore data from manual snapshots.
Load balancing
Alibaba Cloud Elasticsearch distributes incoming requests evenly across all data nodes. Specify the public or internal endpoint of your cluster on your client—the platform handles traffic distribution automatically.
Load distribution depends on the number and size of index shards. Plan your shard count and shard size carefully when creating indexes. For guidance, see Shard evaluation.