[Solved] Slow/hangs ReFs/DPM 2016 with Modern Backup Storage

Back to this issue DPM 2016 UR2 with Modern Backup Storage – server hangs, crashes, backup tasks are slow?

Just got an update from Microsoft Support, the fix for this issue is generally available.

Please use this link to manually download the update – February 22, 2018—KB4077525

Make sure you have added these keys into registry:

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\FileSystem] “RefsProcessedDeleteQueueEntryCountThreshold”=dword:00010000

I didn’t test new update yet, please put your comments, if you have any issues.

You may also like


  1. I have DPM 2016 with rollup 4 installed and windows server 2016 i guess missing latest update. Refs technologie is implemented. I am facing backup jobs slowness and RAM consumption to 95% is this solution is suitable in my case?

    1. Hi Tony!
      In my case I saw few symptoms:
      – RAM usage 95%-99% leads to hangs of the server (ping is OK, but I couldn’t run commands remotely)
      – slow backup tasks (some of them are never ended, especially the sources with large vhdx up to 2-3TB)

      I would suggest you also to check network connection and RAID virtual disk configuration (caching).

      Anyway, applying latest CU mentioned in my article is highly recommended.

      Best regards,

  2. Hi Dmitry!
    I currently have a case open with MSFT Support about this issue. Implemented the KB already + the registry changes, which require a reboot BTW. Nothing helped…
    My backup jobs took 2-3 hours previously, now up to 10-12 hours. Did dome testing with IOMeter on the MBS volume without any DPM jobs running. IOPS start at 2000 and after 10 minutes slowly start to drop. After 1 hour IOPS was on 450, still dropping. When I stop IOMeter there is still a disk queue of 25-50 for more than an hour, which is probably ReFS housekeeping. My conclusion for now in that ReFS is still pretty much unusable for DPM in large environments…
    Don’t know whether it’s possible to use “traditional” DPM storage in DPM 2016 however…. If not, I would have to revert to 2012R2 again.

    Best Regards,

    1. Hi Ad!
      Thanks for feedback.

      I would ask you to check RAID caching configuration on MBS drive (I suppose you have disks in RAID). Try to experiment with small size data sources (less than 1TB), check if you have same symptoms.
      Also check the number of backup tasks. Maybe create few different protection groups so will not run at once.

      You still can use traditional storage in DPM, but your will loose your replicas and they will recreated again. You won’t get all advantages of ReFS like disk space optimization, no LDM DB limit.

      Best regards,

    2. Hi Ad

      Where you ever able to get this issue resolved? I am having exactly the same issue and have already tried the latest fixes like windows updates and registry changes with no luck. I am also thinking of reverting back to 2012 R2.


        1. Hi, Dmitry

          I’m not sure I see the new post about DPM slowness. Can you please provide a link? We are still having issues with slow DPM backups.

          thank you,


          1. Hi Ruben!
            DPM slowness still the issue. There are few important conditions which are must be in place: fast storage system, fast network link and large RAM size. I never seen exact numbers in Microsoft documentation, everything depends on your workload. If you have small amount of data (few TB), probably you will be good with standard server configuration (32 GB RAM, RAID card, 1Gb/s network link).
            I suppose will be some improvements in Win 2019 + System Center 2019.

            Best regards,

    1. Hi!
      The value “00010000” I got from MS Support. It’s also set on my servers, no issues found.

      Best regards,

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.