[DOCS] Guidance on Elasticsearch nodes sizing

[Leaf-Lin]

Today, there's no official public self-service documentation or tool on node size for provisioning new clusters or understanding the resource limitation of existing clusters.

With improved scalability and understanding limitation on shards https://github.com/elastic/elasticsearch/issues/77466, are we in a position to provide a better guidance/calculator for nodes sizing?

There are many tools like this created by 3rd party, for example, https://gbaptista.github.io/elastic-calculator/

Ideally, the official nodes sizing should consider:

Use case:

• Search heavy
  • based on the expected number of users x number of queries x complicity of the aggregation.
• Indexing heavy
  • ingestion data volume
• Security vs Observability vs Enterprise Search

Node types:

• Master
  • based on the total number of indices/field mapping stored in the cluster.
  • does it scale proportional to the number of total nodes in the cluster?
• Data
  • Hot/warm/cold/frozen if retention policy are well-defined.
• Coordinating
• ML

  Snapshot

• Snapshot storage sizing
  • based on retention period and data ingestion
  • if the searchable snapshot is in use, the dimension should include data size.
  • https://github.com/elastic/elasticsearch/issues/67790

[elasticmachine]

Pinging @elastic/es-perf (Team:Performance)

[elasticmachine]

Pinging @elastic/es-distributed (Team:Distributed)

[elasticmachine]

Pinging @elastic/es-docs (Team:Docs)

[henningandersen]

Removing the distributed team labels on this, since most of the guidance for distributed have been done with the many shards project (master and data node heap sizing). I think the bigger picture belongs in the performance team, happy to help out with specifics around it though.