Document scaling guidance for non-allocator hosts in ECE

[rseldner]

It would be helpful to add sizing and scaling guidance for non-allocator ECE hosts and service heap sizes.

We provide a starting point for large installations here: https://www.elastic.co/guide/en/cloud-enterprise/current/ece-install-large-cloud.html
However, there is no follow-up guidance beyond that initial setup.

The Add Capacity guide is allocator-centric and primarily based on the premise of needing to make room for additional deployments: https://www.elastic.co/guide/en/cloud-enterprise/current/ece-add-capacity.html

But what about Directors, Coordinators, Controllers, and Proxies?

We should add documentation covering:

• When to consider scaling up or scaling out non-allocator role hosts
• When it might be appropriate to split the Coordinator role from Director role hosts

Some of the sizing guidance has evolved over time, which can lead to running installations being unintentionally undersized without clear documentation on how to identify or remediate those cases.

A couple examples from large installation configurations where guidance has changed, but no specific upgrade instructions or evaluation criteria are provided in the release notes:

1. Zookeeper and Admin Console heap recommendations were increased from 4GB to 24GB in version 3.6:

• 3.5: https://www.elastic.co/guide/en/cloud-enterprise/3.5/ece-install-large-cloud.html
• 3.6: https://www.elastic.co/guide/en/cloud-enterprise/3.6/ece-install-large-cloud.html

2. Director host RAM recommendations increased from 32GB to 64GB after version 3.1:

• 3.1: https://www.elastic.co/guide/en/cloud-enterprise/3.1/ece-install-large-cloud.html
• 3.2: https://www.elastic.co/guide/en/cloud-enterprise/3.2/ece-install-large-cloud.html