Forum Discussion
Disk throughput performance differences between template and pool VM
Hi,
While testing a Lab I'm preparing, I noticed that the disk throughput performance of my template was as much as 5 times faster compared to a pooled VM from that very same template.
I'm using the "Large VM (nested virtualization)" template with the "128 GB Premium SSD" option.
The template itself reaches about 150 Mbyte/s, while the pool VM maxed out at 30 MByte/s.
The VM is running Ubuntu 22.04 LTS. I'm testing as follows:
dd if=somebigVMfile.vdi of=/dev/null bs=1M status=progressIs there something wrong with the pool VM? Anything it can/should/must do to change this? 30 Mbyte/s is way too slow for a "premium SSD".
Thank you!
I repeated my tests today with the same results. The pool VMs are very slow with disk I/O throughput. I haven't investigated latency yet. Is somebody else having this same problem?
Nobody from the Azure Labs team here to participate?
Stan_AgeevMicrosoft
surinameclubcard Nested virtualization performance is dependent on a number of different factors such as Hyper-V VM settings, Linux OS distro, VHD vs VHDX files, number of nested VMs running, etc. You may want to check this post to see if there's anything you can do to optimize the performance: Optimizing Linux Performance on Hyper-V - Microsoft Community Hub
If that doesn't help, feel free to open an Azure support ticket.
