Windows 2016 Data Deduplication on VMware

[noam126]

Has anyone else tried Windows 2016 native dedup on Nimble volumes?

Partly out of curiosity and partly lured by the appeal of vm-level space savings, I went ahead and enabled Win 2016 dedup on a ~200 GB vm drive. Get-DedupVolume reports SavingsRate ~50%.

There's some discussion of Windows 2012 dedup on another thread and the concerns outlined there would also seem to apply to Win 2016. So far we're not taking Nimble snapshot of this volume and not making CBT-enabled vm backups. We might be incurring some cache-related performance penalty related to additional churn caused by Windows dedup, but that has not caused trouble so far.

Even if our effective capacity is ultimately not improved at the storage array level, (the appearance of) savings at the vm level is nevertheless nice to have for manageability. Admittedly, even the modest cpu & mem overhead could become significant enough to rule out using this feature on a larger scale unless it provides significant capacity savings beyond what Nimble already provides.

Currently there are several dozen vms on the Nimble volume alongside the vm with Win dedup enabled, so I don't have clear picture of the impact of Windows dedup at the Nimble level. I'll post more if I ever get around to doing a more thorough comparison of a Nimble volume with and without Windows dedup. Curious to hear reactions or other relevant experiences meanwhile.

[chris24 · Accepted solution]

Hello, 

As long as you are aware that this should not be used if protecting volumes with a volume collection, much of the space saved within the filesystem will be lost as it will be recorded as change within a snapshot. Installing and enabling Data Deduplication | Microsoft Docs  suggests it's post processing scheduled task, not inline so certainly not something to consider if taking snapshots.

Additionally this would render dedupe on an AFA array mute.

Let us know how you get on

Many thanks,

Chris