VHDX disks can be disconnected during periods of high storage latency

Problem:

FlexApps, ProfileDisks or FSLogix disks exhibit various symptoms during periods of high storage latency on the fileshare hosting these VHDX files.

1. Slow profile or application performance; slow login times compared to non-peak times, for example
2. Windows Event Logs - System - NTFS Read retries (not fatal alone; monitor situation)
3. Windows Event Logs - System - NTFS Delayed write failed (can be fatal; usually accompanied by read retries)
4. Windows Event Logs - VHDMP - VHDX gets unmounted / unsurfaced after too many failures

Discussion:

These technologies rely on Windows' ability to maintain a long-running, stable connection, with minimal latency to remotely-hosted VHDX files.  In our experience, we have found that Windows needs VHDX storage latency to be, on average, better than 40ms.  Brief spikes can be tolerated based on various factors.

These issues can be mitigated by various means, sometimes in-combination, with the goals being to reduce storage latency and increase connection stability.

1. Increase available storage IOPS and/or bandwidth to better handle peak usage scenarios.
2. Consolidate open connections where possible
   1. Related, or "like", applications can be captured in a single package where it makes sense.
   2. Many single-user machines require more open file handles than fewer multi-user machines.
   3. Engage the FlexApp Block Caching feature for a sub-set of package assignments.
   4. Use FlexApp One to sync the applications local, once per app, per machine, before activation.
3. Stagger mounting of FlexApps where possible
   1. Change some FlexApp packages to use the Click-to-Layer option.
   2. Move some FlexApp packages to be activated on-boot.
4. Where supported/applicable, Continuous Availability in SMB3 is required for fail-over support.

Product: ProfileUnity & FlexApp

Product Version: Any

Expires on: n/a

Updated: May 2025