This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

VM Read Rate / Write Rate (MBps) Missing?

Hi all,

We're currently evaluating Solarwinds Virtualization Manager. Primary use case is performance monitoring of our on-premises VMware environment.

We've noticed however that we cannot seem to select virtual disk counters like Read Rate (MBps) and Write Rate (MBps) for each individual vdisk (eg. iscsi 0:0, iscsi 1:0) nor for the total for the VM. From previous troubleshooting events where storage performance was involved we know these metrics to be relevant if not of vital importance.

Have we overlooked something?

Thanks in advance for your feedback/input!

Kind Regards,

Havoc.

  • https://thwack.solarwinds.com/people/havoc%20home  wrote:

    Hi all,

    We're currently evaluating Solarwinds Virtualization Manager. Primary use case is performance monitoring of our on-premises VMware environment.

    We've noticed however that we cannot seem to select virtual disk counters like Read Rate (MBps) and Write Rate (MBps) for each individual vdisk (eg. iscsi 0:0, iscsi 1:0) nor for the total for the VM. From previous troubleshooting events where storage performance was involved we know these metrics to be relevant if not of vital importance.

    Have we overlooked something?

    Thanks in advance for your feedback/input!

    Kind Regards,

    Havoc.

    Hi Havoc!

    I have good news for you, we're actually actively working on adding those metrics in on the VMAN on Orion side. Previously they were available in our appliance deployment and will be ported over to the Orion side in a future release.

    I've indicated this on our roadmap here: WHAT WE'RE WORKING ON FOR VIRTUALIZATION MANAGER (UPDATED MARCH, 2019)

    pastedImage_3.png

    When you're selecting these counters, may I ask what you're planning to do with those counters? Will you be adding them to a PerfStack or alerting? What problem will you be able to solve with access to those metrics?

  • Hi Serena,

    We're constantly monitoring virtual disk Throughput (read/write MBps) to observe extreme behavior. Because Throughput = IOPS * Blocksize, it can have more significance than IOPS alone. Workloads with large blocksizes (eg. databases) usually have more impact than workloads with small blocksizes.

    By using separate Virtual Disk TOP-10s (descending order) for Read Throughput and Write Throughput that refresh every minute we can instantly see when anomalies start occuring and whether they happen for a short or long duration.

    We also distinguish between VDI (Virtual Desktop Infrastructure) and VSI (Virtual Server Infrastructure) by selecting/filtering the correct vCenter instance, so we have separate dashboards (one for our VDI admins and one for our VSI admins).

    Recent real-life troubleshooting events where this was extremely useful:

    - Trashing of the C-drive for a handful (10 out of 1.000) VDI desktops; constant 50MBps write activity during the entire user session. Root cause: Bug in Google Chrome that causes the Chrome db file to be written to many times per second.

    - Trashing of the D-drive for most VDI desktops; constant 10-20MBps write activity to the Windows page file disk during the entire user session. Root cause: Memory depletion.

    - Trashing of the D-drive for one VSI server; constant 80MBps read activity from the data disk, 24x7. Root cause: Scheduled task that got stuck.

    In short: We really need these virtual disk statistics if we're going to remain in-control.

    Kind Regards,

    Havoc.

  • https://thwack.solarwinds.com/people/havoc%20home  wrote:

    Hi Serena,

    We're constantly monitoring virtual disk Throughput (read/write MBps) to observe extreme behavior. Because Throughput = IOPS * Blocksize, it can have more significance than IOPS alone. Workloads with large blocksizes (eg. databases) usually have more impact than workloads with small blocksizes.

    By using separate Virtual Disk TOP-10s (descending order) for Read Throughput and Write Throughput that refresh every minute we can instantly see when anomalies start occuring and whether they happen for a short or long duration.

    We also distinguish between VDI (Virtual Desktop Infrastructure) and VSI (Virtual Server Infrastructure) by selecting/filtering the correct vCenter instance, so we have separate dashboards (one for our VDI admins and one for our VSI admins).

    Recent real-life troubleshooting events where this was extremely useful:

    - Trashing of the C-drive for a handful (10 out of 1.000) VDI desktops; constant 50MBps write activity during the entire user session. Root cause: Bug in Google Chrome that causes the Chrome db file to be written to many times per second.

    - Trashing of the D-drive for most VDI desktops; constant 10-20MBps write activity to the Windows page file disk during the entire user session. Root cause: Memory depletion.

    - Trashing of the D-drive for one VSI server; constant 80MBps read activity from the data disk, 24x7. Root cause: Scheduled task that got stuck.

    In short: We really need these virtual disk statistics if we're going to remain in-control.

    Kind Regards,

    Havoc.

    Havoc, I'm curious after VMAN released virtual disk data in Virtualization Manager 8.5 Release Notes  were you able to set up the custom widgets to get your use case working?