File deleted between snapshots - is data still transfered?

[bepresseditor]

Imagine:

An hourly snapshot happens.

A file is then created.

Then the same file is deleted.

And the next hourly snapshot comes along.

Will the deleted data blocks be part of the next SnapMirror  ® transfer?

I know the file is not part of the snapshot.

Will the data blocks take up SnapMirror bandwidth?

[rkaramchedu1]

What type of SnapMirror? If VSM, then it transfers the snapshots as well so the blocks should make it there.

[bepresseditor]

I'm talking about volume snap mirror here.

The question is really "do the data blocks take up bandwidth".  The file metadata of course is not part of either snapshot.

[ekashpureff]

If it's sync or semi sync snapmirror, then the file will take up bandwidth.

If it's async snapmirror and the file existed at the time of a transfer, then it wil take up bandwidth.

I hope this response has been helpful to you.

At your service,

Eugene E. Kashpureff
ekashp@kashpureff.org
Fastlane NetApp Instructor and Independent Consultant
http://www.linkedin.com/in/eugenekashpureff http://www.kashpureff.org/

(P.S. I appreciate points for helpful or correct answers.)

[bepresseditor]

It is asynchronous snapmirror.  The snapshot schedule is daily.

The files in question are temporary: currently created and then just left around.

They'll snapmirror for sure.

If we change the system to delete the files, other than the few that happen to cross a time boundary,

we're hoping for a bandwidth usage reduction.

Since snapmirror works at the block level, we were unsure if deleting the files would make any difference.

If snapmirror works off the block lists in the file metadata, we're fine.  If it works at a true block level,

likely we'd get no benefit from the change, as the blocks would come over anyway.

[ekashpureff]

My understanding is that it is block for block, including all metadata and snapshots.

Deleting the files should reduce the transfer bandwidth, as you have proposed.

I hope this response has been helpful to you.

At your service,

Eugene E. Kashpureff
ekashp@kashpureff.org
Fastlane NetApp Instructor and Independent Consultant
http://www.linkedin.com/in/eugenekashpureff http://www.kashpureff.org/

(P.S. I appreciate points for helpful or correct answers.)

[aborzenkov]

There is a corner case – while free blocks are not transferred, freeing blocks is not instantaneous. For a large file it may take significant time. So if SM was initiated while free block scan is not yet complete, blocks not yet marked as free may be transferred.

[ekashpureff]

No, the file will not be replicated. It was never versioned in a snapshot.

I hope this response has been helpful to you.

At your service,

Eugene E. Kashpureff
ekashp@kashpureff.org
Fastlane NetApp Instructor and Independent Consultant
http://www.linkedin.com/in/eugenekashpureff http://www.kashpureff.org/

(P.S. I appreciate points for helpful or correct answers.)

[rkaramchedu1]

Doh! - should have read the post more carefully.

Eugene is on point. If the file is created and deleted between snapshots, then it is not captured.