Hi, my name is Eddy Kim.
I'm not good at English. Please look at this.
We're going to build our own ElasticSearch infrastructure, and we're going to buy a server soon.
Our systems are not big data, but indexing performance is very important. I'm going to make 3 to 5 nodes. I want to know the pros and cons of four methods. I don't want to be overly engineering.
I don't know if I need a client node.
Why client nodes are typically required?
If it's true that we don't need a client node...
I'm looking at number1 and number4.
Number1 and number4...
I'm not sure if the indexing performance is very different.
Anyway, for a little bit of faster performance, I'd like to make a final decision with Number1, which is separated from Master Node, but I'm not sure if it's a valid decision.
Indexing performance will likely be limited by the data nodes so adding all the other node types may not give you much. You should also always aim to have at least 3 master eligible nodes which a few options do not. Adding 3 dedicated master nodes may be good down the line if your master/data nodes are under very heavy load, but you should then not send requests through them and leave them to manage the cluster.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.