How can we improve Azure Virtual Machines?

Storage Optimized VMs should be having higher IOPS and MB/s throughput

Currently Storage Optimized VMs have very limited IOPS and MB/s throughput.

Our Oracle server has 6x1TB premium (6xP30) disks added to it. Which can go upto 1200MB/s throughput. Disk Block Sizes are configured at 1MB. Still we are restricted to the VM limit of 125MB/s. (PFA the current options)

Please consider higher VM level limits.

11 votes
Sign in
Check!
(thinking…)
Reset
or sign in with
  • facebook
  • google
    Password icon
    Signed in as (Sign out)

    We’ll send you updates on this idea

    Jenson Joseph shared this idea  ·   ·  Flag idea as inappropriate…  ·  Admin →

    1 comment

    Sign in
    Check!
    (thinking…)
    Reset
    or sign in with
    • facebook
    • google
      Password icon
      Signed in as (Sign out)
      Submitting...
      • TJ Cornish commented  ·   ·  Flag as inappropriate

        We agree, and total VM throughput is hard to find in the documentation. The VM selection screen proudly shows "8000 IOPS" or whatever, but the reality at least for our general computing tasks is we are never even close to the IOPS limit, but we are limited by the VM throughput limit.

        This is especially frustrating in that a single transfer between drives on the machine costs twice - i.e. copying from the E drive to the F drive means your throughput is only half what is advertised as the machine capability.

        Machines like the E4s_v3 have 4 CPUs and relatively a lot of RAM - 32GB compared to a typical client machine (desktop or laptop), but the VM throughput limit is SO BAD compared to the experience on a typical client machine. It is very frustrating to have an application that runs faster on a $1000 desktop computer than on a $600/month Azure VM. Increasing to the next VM size gives us cores and RAM we don't need, and now we're talking $1000/month to have comparable performance to a $1000 to purchase desktop.

        I just found out about the new "L-series" storage-optimized VMs, which unfortunately aren't available in our region yet, but even these have a throughput limit that is only 33% faster than the E series (96MB/sec for E4s_V3 vs 125MB/sec for L4s). Limits across the board - A series, D series, E series, etc., should be 2X greater or more, and "storage optimized" machines should be faster yet.

      Feedback and Knowledge Base