Announcing Public Preview: FSLogix Disk Compaction

I would think that if the profile is mounted 100% of the time because the user isnt logged off, that you could not compact the profile properly anyway using a script. Perhaps it would be easier to configure the maximum time that compaction can happen during log off to something less than 5 minutes, if you think users will be sensitive to that? Seems like a non issue.

[NK] - Interesting feature, but how it helps in user's experience? I think running your great compact script offline during maintenance window is sufficient than increasing the log out time for the user, but as you stated the process will time-out if it takes longer than 5 minutes and continue from the previous state - Ok but still increasing log out time for the user for the "first time", BUT the compact operation logic will still need to evaluate as to know if compact is needed and how much time to compact.

However, the question stands still - benefits of allowing FSLogix Compact to invoke during log off Vs Running your compact script offline once a in week during maintenance window is something we need to evaluate.

Do they Log off? Consider this situation - you are saying that the FSLogix compact operation will run only at user log off. However, not all users LOG OFF off their systems. That means if the user doesn't log off then the FSLogix Compact operation will not run? Is this correct? If yes, then there are chances that only 20% of users or less will be touched by FSLogix Compact Operation and rest will be untouched. On the other hand, if you run your script offline once in a week, we are ensuring to target all users.

Profile Sizes: Secondly, evaluate how many users you have in your environment and what's their "Average" size of profiles - In an environment where the profile size is lesser than 4-5GB GO ahead and enable it as it won't take much time (it wouldn't even take 1 minute as those compact commands are faster) and since FSLogix uses APIs (instead of using PowerShell commands) the process is quite faster.

Run Offline Script Once: Finally, If you really want to enable FSLogix Compact Operation at log off then before enabling always run the script offline ONCE to ensure white space from all production profiles have been removed - This way compact operation logic doesn't have to spent much time in figuring out how much data blah blah need to be adjusted/compacted or whatever. 

Any Controls? Moreover, not sure if there is any registry entry that will control compact operation at host level (yes, I guess) as well as user level (I guess no) or be able to run Compact command only if the user profile size is more than "x" GB (I guess no).

If no additional controls in place, then only enable it for specific hosts (let’s say 10 hosts at a time) to see impact (basically user experience).

Best.

I'm not seeing the download for 2210 yet. Do you know when it will be available? Thanks.  (Seemed to be posted shortly after this post so disregard - Thx)

The I/O load is minimal as we don't actually move any blocks around, it's more a filesystem metadata activity, so you will see a very small uptick in CPU on the host and on the disk controllers for whatever storage you are using, but that's it.

Good point Travis_78 How much I/O should we in average expect here. Do you have any numbers on this JimMoyle? 

This is great news. It surprised me that this was not part of the solution all along. Although, in huge deployments, you wonder what the disk I/O would be like on the first run at the end of the day when many people log off at once..