How to convert hot induce to cold/warm/hot plus hardware profile change?

Hey all,

I’m using Elastic Cloud.

We have a large data (900+GB)set that we’ve indexed as hot storage on IO hardware profile.

This is naturally expensive.

What id like to do is two fold:

Change the IO profile to Compute.

Change the way we have the induce setup - I’d like to store the entire dataset as cold storage, then automate warm and hot based on query frequency.

I’m an amateur here so I’d appreciate any guidance with how we can achieve the above, please.


Howdy! Since you're using Elastic Cloud, you may consider filing a case to Elastic Support instead of generic Elastic Discuss community members, so that the team can investigate your direct deployment & offer direct/data-sensitive advice privately :slight_smile:

To your questions, generically

  • To change hardware profiles: you'll need to restore snapshot into a new cluster of the desired hardware profile
  • Change induced setup: I recommend against Prod testing & also against shifting everything to cold nodes and then by query frequency moving indices back. Usually when adding data tiers, (spitballing game plan) users will
    1. investigate index query frequency (e.g. via Index stats)
    2. add ILM policies to indices/templates (e.g. blog & blog)
    3. Expand the rollover/min_age values on ILM policies associated to index/template per high search volume/frequency & start pushing less accessed data into warm and then cold nodes
1 Like

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.