On 18/02/2022 18:41, Melanie Plageman (Microsoft) wrote:
For example: a device provisioned with high IOPS and BW limits on the same controller as a smaller and slower device can starve the slower device of tags. This is especially noticeable when the slower device's workload has low I/O depth tasks.
If you check hctx_may_queue() in the block code then it is noticeable that a fair allocation of HW queue depth is allocated per request queue to ensure we don't get starvation.
Thanks, John
