Skip to main content
Version: User Guides (Cloud)

Manage Replica

Zilliz Cloud supports cluster-level replication. Each replica is an exact copy of the resources and data in a cluster. Using replicas can increase query throughput and availability.

For users with small datasets experiencing QPS bottlenecks, adding replicas can distribute the query workload, enhancing overall query throughput. However, adding replicas will not increase the cluster capacity because the capacity is only determined by the CU size of each cluster. If you want to increase the cluster capacity, please refer to Scale Cluster.

Configuring replicas will affect the cluster's monthly CU cost. The storage cost of the cluster will remain unchanged. For more information, refer to Estimate Cost.

This guide outlines the procedures of configuring replicas for a cluster in Zilliz Cloud.

📘Notes

This feature is currently exclusively available to Dedicated (Enterprise) clusters.

Limits​

You can configure replicas for an existing Dedicated cluster as long as the following conditions are met:

  • The cluster has 8 CUs or more

  • For clusters whose compatible Milvus version is lower than 2.4.13, all collections in the cluster need to be released

Note that the product of the cluster CU size x replica count should not exceed 256.

🚧Warning

Updating the replica configurations may lead to slight service jitter. Please exercise caution.

Configure replicas manually​

You can adjust the number of replicas for an existing Dedicated cluster either manually on the console or programmatically.

Via web console​

The following demo shows how to configure replicas on the Zilliz Cloud web console.

Via RESTful API​

You can use the RESTful API to manually adjust the number of replicas in a cluster.

The following example manually sets the cluster replica count to 2. Note that the value for the replica parameter should be an integer ranging from 1 to 8. For details, refer to Modify Cluster Replica.

export BASE_URL="https://api.cloud.zilliz.com"
export CLUSTER_ID="YOUR_CLUSTER_ID"
export TOKEN="YOUR_API_KEY"

curl --request POST \
     --url "${BASE_URL}/v2/clusters/${CLUSTER_ID}/modifyReplica" \
     --header "Authorization: Bearer ${TOKEN}" \
     --header "Accept: application/json" \
     --header "Content-type: application/json" \
     --data-raw '{
        "replica": "2"
      }'

Auto-scale replicas​

Currently, you can only auto-scales replicas based on a predefined time schedule via the Zilliz Cloud web console.

The following demo shows how to enable replica auto-scaling.