OpenSearch sharding for 3 nodes cluster - Recommendations

OpenSearch
1

Versions (relevant - OpenSearch/Dashboard/Server OS/Browser): 2.19.2

Describe the issue:
Need the recommendation for data sharding for 3 node cluster, Reg : no of primary shards ? and no of replicas ? ..
Need some one advise the best approach to provide the primary shards/replicas to maintain for 3 node active cluster and approach …

Configuration: opensearch.yml

Relevant Logs or Screenshots:

2

Hi @vinay ,

It’s hard to say an exact setup which will be best as this depends on many variables. I would first figure out a few different questions to better determinate a possible solution.

  • How many index’s will it consist of?
  • How big will the nodes be? CPU, Ram, How much heap size? However stuff like this will be better determined with time and usage.
  • How much data will you be storing per index?
  • How long will you retain that data for?
  • What will the data be used for? Searching? Faster indexing?

For now this should be enough to go on.

You can read the below doc to get an idea on shards:

Personally I would start with 1 index and 1 replica not to lose data. I would also look into Data Streams Data streams - OpenSearch Documentation and ISM Index State Management - OpenSearch Documentation to help manage your setup.

Leeroy.