Steps to reproduce:

Take any elasticsearch cluster with a large amount of nodes and apply Zabbix elasticsearch template

Result:

Template returns very large amount of information. Most of this data is NOT used in template. For large elastic searh clusters size of date may reach tens of megabytes. When customer have several such large clusters this overhead of data significantly reduce zabbix performance.

Needed:

Optimization of eleasticsearch template.
Make template return less data , and filter out only required data by using filter_path and other methods. Data may be reduced by order of magnitude. 

For example for node discovery is reqest

GET _nodes/_all 

is used  and it return 117491 bytes for three node cluster.

by using filter **

GET /_nodes?filter_path=nodes.*.name
or
GET /_nodes?filter_path=nodes.*.name,nodes.*.ip,nodes.*.version

this size is reduced to 134 bytes.  

(it was tested on three node ES cluster v8.18.3 deployed in docker https://www.elastic.co/docs/deploy-manage/deploy/self-managed/install-elasticsearch-docker-compose)