Coordinating only nodes act as smart load balancers. This special role of the coordinator-only node provides advantages for a wide range of clusters by offloading the coordination responsibilities of the data and master nodes. After joining the cluster, these nodes, like any other node, obtain the complete cluster state and use this information to efficiently direct requests to the appropriate destination.
Adding coordinator nodes to an Elasticsearch cluster can be a strategic move to improve cluster performance and efficiency. Coordinating nodes, also known as client nodes, are essentially Elasticsearch nodes that do not save any data or perform any calculations. Their main role is to route search and indexing requests from clients to the appropriate data nodes and then aggregate the responses.
Coordinating node
Requests such as search requests or batch indexing requests may involve data held on different data nodes. For example, a search request is executed in two phases, which are coordinated by the node that receives the client request (the coordinating node). For information on how data is indexed, read the article " Elasticsearch: How Indexing Data is Done ". For information on how the data is read, read the article " Elasticsearch: How is the data read? ".
During the decentralization phase, the coordinating node forwards the request to the data nodes that hold the data. Each data node executes the request locally and returns its results to the coordinating node. During the collection phase, the coordinating node reduces the results of each data node into a single global result set.
Each node is an implicit coordinating node. This means that nodes with an explicit empty role list via node.roles will only act as coordinating nodes and cannot be disabled. Therefore, such a node needs to have enough memory and CPU to handle the collection phase.
Detailed reading: Node | Elasticsearch Guide [8.10] | Elastic
Here are some scenarios where you should consider adding a coordinator node to your cluster
- High query load : If your cluster experiences high query load, adding coordinator nodes can help distribute the load more evenly. Coordinating nodes can handle the task of distributing queries and aggregating results, freeing data nodes to focus on executing queries.
- Complex aggregations : If your use case involves complex aggregations, a coordinator node may be useful. Aggregation can be resource intensive and offloading this work to the coordinating node can help improve performance.
- Large number of indexes or shards : If your cluster has a large number of indexes or shards, the task of routing requests can become quite complex. Adding a coordinating node can help manage this complexity and make request routing more efficient.
- High client connection count : If you have a large number of clients connecting to the cluster, adding a coordinator node can help manage those connections more efficiently. The coordinator node can handle client connections, freeing the data nodes to focus on processing requests.
- Hybrid cloud or multi-region deployment : In a hybrid cloud or multi-region deployment, the coordinator node can be used to route requests to the appropriate data nodes based on factors such as data location and network latency.
Before adding a coordinator node, you must consider the impact on cluster resources. Coordinating nodes require CPU, memory, and network bandwidth, so you must ensure that the cluster has sufficient resources to support the other nodes.
Also, keep in mind that adding coordinator nodes is not a panacea for all performance issues. It is important to monitor the performance of your cluster and make adjustments as needed.
Here are some questions and answers
Q: What is the coordinator node in Elasticsearch?
Answer: A coordinator node, also known as a client node, is a node that does not save any data or perform any calculations. Its main role is to route search and indexing requests from clients to the appropriate data nodes and then aggregate the responses.
Q: When should I consider adding a coordinator node to my cluster?
Answer: If your cluster is experiencing high query load, if your use case involves complex aggregations, if your cluster has a large number of indexes or shards, if you have a large number of clients connecting to your cluster, you should consider adding a coordinator nodes, or if you're running a hybrid cloud or multi-region deployment.
Q: How does the coordinator node improve the performance of the cluster?
Answer: The coordinator node can improve performance by offloading the task of routing requests and aggregating responses from data nodes. This allows data nodes to focus on executing queries, resulting in faster response times.
Q: What resources are required for the coordination node?
Answer: Coordinator nodes require less DISK, CPU and RAM than data nodes.
Q: After adding how many nodes should I consider adding coordinator nodes to my Elasticsearch cluster?
A: The decision to add coordinator nodes is not strictly based on the number of nodes in the cluster. It's more about the workload and performance of the cluster. If your cluster experiences high query loads, complex aggregations, large numbers of indexes or shards, or large client connection counts, adding a coordinator node can be helpful. However, as a general guideline, when your cluster grows beyond 10-20 nodes, you might start thinking about adding coordinator nodes. Remember, it's important to monitor the performance of your cluster and make adjustments as needed.