Exchange best practices with ReFS repo

[pj888888888888]

So I am just trying to get a feel of what the best settings for exchange DAG backups would be.
RIght now we are getting pretty decent backup times on our exchange environment (Usually the incrementals take 10-20 minutes)
But when our fulls run, the synthetic full process can take up 48+ hours due to the transformation process....and I'm curious if its even necessary to convert the increments to rollbacks (being we have multiple runs during the day to control the log growth)
Anyone have any tips or suggestions?

[Delo123]

ReFS should be very quick on synthetic fulls, or are you doing reverse incrementals maybe, if so, why?

[Mike Resseler]

Guido is correct.

What are your settings? doing synthetic fulls on ReFS 2016 should give you massive time savings.

[pj888888888888]

So I am doing traditional incremental (not reverse), but the synthetics are just bad (They take up to 24 to 34 hours)
My current jobs settings are set to do Synthetics every weekend, I had another synthetic set up during the week, but it would run over 30 hours at times and was impacting how often the logs were being truncated.
I changed the synthetics to not transform the previous chains into rollbacks, but I wasn't sure if that was a bad idea. (It appears to have helped, it only took 20 hours to conduct the synthetic)

I was more so seeing what everyone else was doing/what the best option would be.

[Mike Resseler]

Paul,

I might have misread it, are you using Win2016 ReFS on your repositories?
In any case, it might be interesting to log a support call (post the Case ID and outcome here) so our engineers can see whether this is expected or not (or where there could be a bottleneck)
Thanks
Mike

[veremin]

I'm wondering whether you're leveraging ReFS 3.1 (Windows Server 2016) at the moment or not. If yes, are you sure the corresponding functionality has taken place indeed (check the job statistics for [fast clone] string)? If not, are positive that at least one active full backup has been executed since you introduced ReFS 3.1 repository?

[pj888888888888]

Yeah its 2016 using ReFS 3.1...it always tends to blame the disk being we are leveraging our FC SAN....and honestly...its just the synthetics that run slow.
We did do all the steps above, we migrated the backups over and did an active full, and I do see the fast clone activity taking place during the synthetics.

I may submit a support call, under a Sev 4 though....being its not really broken, it just tends to drag....and it is a 8TB VM.
Honestly....its still better than before, it would take 30+ hours to do an incremental....I'm just being a perfectionist

[Mike Resseler]

Paul,

Please do. Our support engineers will be able to look into the logs and see if there is something, or that this is what you can expect.

If you have a support case, please post it here and the result of the support call after that
Thanks
Mike

[pj888888888888]

Mike,

Will do, there might be a little delay, we are in the process of renewing our licensing and we are waiting for the new ones to be issued (you know how that goes)
I'll more than likely submit this as a Sev4 being its not really down, I just want it to be faster.