MemberMay 30, 2020 at 5:49 pm
Thank you for your feedback.
The queue length is the number of read or write operations waiting to be completed on an EBS volume. When you use Amazon EBS Provisioned IOPS volumes, you will want to deliver optimal levels of input/output performance based on your workload.
You need to match the EBS volume workload to the IOPS you provisioned. If the IOPS demand is higher than what you provisioned and the application is latency sensitive, consider using a Provisioned IOPS (SSD) volume with more provisioned IOPS. If I/O latency is high, you can check the average queue length to be sure that your application is not trying to drive more IOPS than you provisioned.
On this question, the company is using an HPC cluster that has transaction-intensive, low-latency workloads. So adjusting the volume’s provisioned IOPS based on the workload makes sense.
For computation, any volume 640 GiB in size or greater allows provisioning up to a maximum of 32,000 IOPS. Therefore, (32000/640) = 50 IOPS/GB. So for the 10GB volume, and you want to maximize IOPS, you can set it to 500 IOPS (50 IOPS x 10GB).
Hope this helps.
Kenneth Samonte @ Tutorials Dojo