Zone/Node/Index Allocation


(David Koblas) #1

I've got a fairly large document set (~1B) and want to organize things in a
performant fashion with enough replication to make me happy.

The general configuration:
3 indexes
50 shards
X nodes - all AWS

What I would like to do is configure this using the cluster zone/nodes in
such a way that I can have:

N - persistent nodes - contain the whole collection of docs (fewer
nodes, slower, more storage)
M - query nodes - insure that only one replica is on each of these nodes

I'm pretty sure that this is do-able with the zones/nodes but didn't quite
figure it out.

Is the following correct:

Set :
node.rack_id : persist
node.group: persist
and/or
node.rack_id : query
node.group: query

Then set:
/_settings
index.routing.allocation.include.group1 : persist
index.routing.allocation.include.group1 : query
cluster.routing.allocation.awareness.attributes : rack_id

Does the above work, before I try it?

If there is anything else in terms of advice that would be useful?

Thanks,
David

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


(system) #2