Hey Checkyourlogs Fans,
In today’s post, we will look at the steps required to enable deduplication in Windows Server 2019 on a ReFS volume. What I have done to get started is installed a Virtual Machine running Windows Server 2019 Insiders Build.
Next, we need to install the deduplication feature as seen in the image below.
Once installed you can use Server Manager to configure it on your Volume. In this example, I have added an E: drive and formatted it using ReFS.
To enable deduplication on the ReFS volume you simply need to right-click on the volume and click Configure Data Deduplication.
This will open the Deduplication Settings page and now we just need to make a few configurations and we are done.
For this use case, I am planning on using the volume for my Hyper-V Lab Virtual Machines.
I have changed the Data deduplication setting to Virtual Desktop Infrastructure (VDI) Server and Deduplication files older than (in days): 0.
The last step is to configure the schedule and we are done. Click on the Set Deduplication Schedule button.
In my lab,, I always choose all three options and leave the default schedule. It is really important to note that just like Deduplication on an NTFS volume this is all done post-process. Meaning that you have to wait for the scheduled tasks to run for the Deduplication engine to start working.
We can check the status using PowerShell which I will cover in the next blog post. For now, I will leave you with how to manually kick off using the build in scheduled tasks.
I really hope you enjoyed this post and stay tuned for a lot more on Deduplication on ReFS.
Thanks,
Dave
hi Dave, thanks for your blogpost!
I get an awful performance with dedubliction enabled, about 4x times slower.
Do you get the same?
Happy New Year and Merry Christmas, by the way 🙂
You will se a bit of a slowdown –> How do you have your volumes configured –> Mirror / 2way / 3Way Parity
hi Dave, sorry for late reply.
My config is Two-Way-Mirror out of 4xSSDs.
I would add a few SSD’s and run 3 way mirroring
Dave what would be the CPU/Memory overhead impact for enabling dedup on an S2D cluster. Any testing done for the same ?
Yes there is a hit of about 20 % host memory + CPU –> Make sure to run Dedup during off peak hours if possible or change the schedule to be less aggressive.
Check my latest blog post I did some testing on this.