Help Center/ Cloud Search Service/ User Guide (Paris and Amsterdam Regions)/ Scaling In/Out a Cluster/ Overview
Updated on 2023-06-20 GMT+08:00
View PDF
Share

Overview

You can scale in or out a cluster and change cluster specifications. In this way, you can improve cluster efficiency and reduce O&M costs.

Scaling Out a Cluster
  • If a data node (ess) processes many data writing and querying requests and responds slowly, you can expand its storage capacity to improve its efficiency. If some nodes turn unavailable due to the excessive data volume or misoperations, you can add new nodes to ensure the cluster availability.
  • Cold data nodes (ess-cold) are used to share the workload of data nodes. To prevent cold data loss, you can expand the storage capacity of the cold data node or add new ones.
Changing Specifications
  • If the allocation of new indexes or shards takes too long or the node coordination and scheduling are inefficient, you can change the master node (ess-master) specifications.
  • If too many tasks need to be distributed or too many results have been aggregated, you can change the client node (ess-client) specifications.
  • If the writing and query of a node suddenly become slow, you can change the data node (ess) specifications.
  • If cold data query becomes slow, you can change the cold node (ess-cold) specifications.
Scaling in a Cluster
  • If a cluster can process existing data without fully using its resources, you can scale in the cluster to reduce costs.
Removing Specified Nodes
  • If a cluster can process existing data without fully using its nodes, you can remove one or more specified nodes from the cluster to reduce costs.
Parent topic: Scaling In/Out a Cluster