File pool policy - Impact of 'Move Snapshots to storage pool or tier' setting

Question

Can someone explain to me the impact or behavior of the setting "Move Snapshots to storage pool or tier" in a certain scenario?

We plan to add F200 nodes "ontop" of our H500 nodes and use file pool policies to start tiering. Some datasets shall be hard fixed in the SSD tier, some in the HDD tier and the rest (default) "tier by age/last modified" from SDD to HDD.

How do existing/new snapshots affect file policy data movement from tier 1 to tier 2 and vice versa? We rely heavily on GFS-style snapshot retention on the data (for "previous versions" restore purpose) that would be affected by the "tiering by age/last modified" policy.

Suppose we have existing snapshots on files/data that currently reside in the SSD tier, and these snapshots will age out and be deleted AFTER the files/data would be moved to HDD by file pool "tiering by age/last modified" policy.

What is the effect? And how does the file pool policy setting "Move Snapshots to storage pool or tier" affect this? Will the files/data and the snapshot "space" move between tiers according to the file pool policy? Or will the files/data and the snapshot "space" that would move according to the file "tiering by age/last modified" policy stay in the SSD tier and only move and free up SSD space AFTER the Snapshots expired and have been deleted?

The documentation I found is not really getting into this. Maybe someone can explain from first hand experience...?

Thanks in advance for all helpful input!

Phil.Lam · Answer

@CendresMetaux

How do existing/new snapshots affect file policy data movement from tier 1 to tier 2 and vice versa? We rely heavily on GFS-style snapshot retention on the data (for "previous versions" restore purpose) that would be affected by the "tiering by age/last modified" policy.

PHIL>Point snapshot to H500 so it won't take space on F200. The default is if snapshots are taken on F200, they will use F200 pool. It should not affect any "previous versions", just where it stores snapshots. Snapshots doesn't move when tiering is involved.

HTH
http://www.unstructureddatatips.com/isilon-quick-tips-compare-snapshots-in-onefs/

CendresMetaux · Answer

Hi Phil

Thanks for jumping in!

Does this mean that the following data model is correct:

1. New File "A" gets stored with Compression+In-Line Dedupe on F200 pool

2. Snapshot gets created on this file by schedule (with move snapshot to H500 pool configured) -> no additional space usage yet on any tier (as the new snapshot has "0" size for the moment)

3. File "A" gets modified -> changed bits will be written/stored in H500 pool "in the snapshot" without Compression/Dedupe as this pool does not support Compression+In-Line Dedupe.

If above is correct, then the following statement is also correct, right:

A) File "A" gets "Bit-fragmented" between F200 and H500 pools as long as 1) snapshots exist "on" it and/or 2) it has not been moved to H500 pool by filepolicy

Phil.Lam · Answer

@CendresMetaux ,   yes on 1, 2, 3.

Isilon

File pool policy - Impact of "Move Snapshots to storage pool or tier" setting

Was this post helpful?