StarWind Software

So I have a file server, it has two VHD files, C drive and D drive. Previously the VM was running on a non-clustered hyper-v host, ran fine, no issues.

Moved it to the starwind native san hyper-v cluster, and now the D drive is super slow.....on the old box it was on 1 gbps connection and I could max it out.

Now it's on a 10gig connection and I get about 20MB/s out of it, it peaks around 40 but slows down to 20 in no time. When I copy files from the C drive....they max out the client nic of 1gbps.

C and D drive VHD's are on same starwind target..........

Weird.....

Dynamic VHD files and heavy fragmentation?

Actually, the d drive is a fixed size and the c is dynamic...and the c drive is faster!

It makes absolutely no sense whatsoever. They are both on the same starwind target, even benchmarks show the d drive to be abysmally slow compared to the c drive.....

I may just rebuild the VM and see if it still does it....

Interesting. Can you run a benchmark Vs. raw StarWind volume?

anton (staff) wrote:Interesting. Can you run a benchmark Vs. raw StarWind volume?

Well..not really since its currently mounted as CSV.

But other VM's on the target operate fine, even with dynamic vhd's, so I'm leaning towards a problem with that particular vhd, or the fact that its a second vhd attached to the vm (all others are single vhd attached vm's), and less likely to be starwind itself.

So weird, i can get 60k iops from other VM's but this one vhd can't break 400.......

Can you confirm that you are using GPT on all NTFS layers?