top of page

Calculating Disk Performance

Calculating Disk performance and Latency

In this article, we shall discuss on performance calculations on disk and storage latency under virtual environment. Overall subject can be segregate into three areas.

  • Poor virtual machine performance

  • Poor disk latency during backups

  • Virtual disk latency

​Above counters may occur due to multiple reasons and one of major cause is, your virtual machines do not have enough I/O per second (IOPS) and IOPS available per virtual machine is less than 30. Now, points is how do i calculate IOPS per Virtual machines ?. Each disk in the RAID volume backing your LUNs is capable of a certain amount of throughput. For a 7200 RPM (7k) drive, you get approximately 100 IOPS per disk, and the I/O per disk increases as the spindle speed increases. The IOPS per disk vary depending on the RAID type used when configuring your array and/or LUN. For more information, contact your storage vendor. ​These numbers indicate the IOPS that is possible using the VMFS file system, which is optimized to avoid hot spots on the disk and is not optimized for raw access speed: Disk Spindle Speed (RPM) IOPS available

  1. 7200 RPM100

  2. 10,000 RPM150

  3. 15,000 RPM230

Calculating the total IOPS per virtual machine

To calculate the total IOPS per virtual machine, find the speed of your drives and determine their IOPS. Each drive in your RAID volume within your SAN contributes to the total I/O available for that volume. IOPS per virtual machine is calculated by dividing the total capacity available by the number of virtual machines running on that volume (which may contain more than one LUN). For example, if you have six 10,000 RPM drives, the total IOPS available in that volume is 150 x 6 = 900. If you are running 50 virtual machines on LUNs located on this RAID volume, each virtual machine gets an average of 900 / 50 = 18 IOPS. In this case, you experience relatively poor virtual machine performance. In this example, to obtain acceptable performance you should not exceed 900 / 30 = 30 virtual machines on that volume.

Note : Backing up the storage consumes more IOPS and adds an additional strain on the volume. In this case, you may want to have an additional overhead capacity to handle this. To determine the amount of additional capacity you may require to perform backups, contact your backup vendor. also, hyperthreading does not actually double the available pCPU. Hyperthreading works by providing a second execution thread to a processor core. When one thread is idle or waiting, the other thread can execute instructions. This can increase efficiency if there is enough CPU Idle time to provide for scheduling two threads

​Best Practice considerations​

  • Start with one vCPU per VM and increase as needed.

  • Do not allocate more vCPUs than needed to a VM as this can unnecessarily limit resource availability for other VMs and increase CPU Ready wait time.

  • A general guide for performance of {allocated vCPUs}:{total vCPU} from the Best Practices recommendations is:​

1:1 to 3:1 is no problem

3:1 to 5:1 may begin to cause performance degradation

6:1 or greater is often going to cause a problem

Right sizing in environment will help you to avoid last movement issues under your infra and increase overall infra uptime.

​For more about Disk performance and types, read my another article which associated with types of storage disk.

4 views0 comments


Post: Blog2_Post
bottom of page