Size index


(Fram Souza) #1

Hello, I'm working on an Elasticsearch implementation project, and I'd like to know the best scenario for mapping the indexes.
Will be imported within the index 130 GB of data that are necessary for operation, after import, we will have the increment of 2GB monthly within that same index.
What would be the best mapping?
1 - Monthly indexes even if the size of shards are small?
2 - Annual indices since index size is relatively small?
3 - Increase the amount of shards?

The profile is 80% read and 20% write. I also have 5 nodes in the cluster, Should each node contain a shard?


(Thiago Souza) #2

If indexing rate is low then there is no need to scale the numbers of shards to the nodes of your cluster (increasing shards is needed when more indexing throughput is required).

Also, we recommend to keep shard sizes around 50Gb and, according to what you have described, I suggest that you take a look in the Rollover Index feature as it gives more flexibility beyond time-based indices.


(Fram Souza) #3

Hi @thiago ,

I'm use API rollover and curator for handle index and snapshots. As the index rate is low I kept 5 shards and 1 replica.
Thanks for reply :wink:


(Thiago Souza) #4

@Fram_Souza if indexing rate is low (like <1k/s) then you could easily stay with 1 shard only (as long as you have them under 50GB) and use the cluster more for replicas, which seems is more needed (due to higher query rate)


(system) #5

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.