What is the typical way to scale out the coordinating-only-node?

[YuWatanabe]

Hello.

I would like to get advice for setting up multiple coordinating only nodes when using with kibana .
I have read below document in advance of asking this question.

https://www.elastic.co/guide/en/kibana/6.2/production.html#load-balancing

I am assuming case where you use coordinating only node for cross cluster search client as well . Cases when increasing nodes will be large heap usage because of increased amount of user sessions and aggregations .

Do you usually add load balancer in front of kibana and increase the set of kibana and coordinatin-only-node , or add loadbalancer beneath kibana and add multiple coordinating-only-node ?

Any recommended architecture for setting multiple coordinating-only-nodes?

[Nathan_Reese]

The usual architecture is to add load balancer in front of kibana and increase the set of kibana and coordinatin-only-node.

[YuWatanabe]

@Nathan_Reese

Thank you for the reply.
Looks cool . I will proceed testing in my lab.

If my understanding is correct, both kibana will point to the same index (kibana.index) which is .kibana. Am I correct?

https://www.elastic.co/guide/en/kibana/current/settings.html

Also , for future reference , I found the perfect blog which describes reply from @Nathan_Reese .

Kibana High Availability

Scaling Elasticsearch, Kibana, Beats, and Logstash

[Nathan_Reese]

Yes, all kibana instances should point to the same kibana.index