"client node" in AWS Elasticsearch?

2
votes

Reading about the proper architecture for Elasticsearch, I hear about 3 types of nodes:

  • Master node (Keeps the shards routing table)
  • Client node (Does the CPU/Memory intensive merge sort of the results)
  • Data node (Contains the actual shards with the data)

From my understanding when connecting to the Elasticsearch cluster using non Java clients, you should be connecting to the client node. However I don't see anything about the Client Node on AWS Elasticsearch.

Am I missing something? Thanks, Sep

1
amazon-web-serviceselasticsearchload-balancing
I believe the java client you use to connect to the cluster is the one called by the name "client node" – Rajind Ruparathna

1 Answers

5
votes

The Amazon Elasticsearch Service is not a full-featured ES installment. Even though, it is sufficiently featured for a few use cases (log collection, etc), it has several limitations in terms of what you can configure and which endpoints are available.

One limitation is that you can only configure master and data nodes, no client nodes.

Another limitation is that the only way to connect to the cluster is through HTTP, i.e. there is no way to get a TCP connection to the cluster, hence you cannot use the transport client, only HTTP clients.