How the flush efficiently the disk cache of a hybrid Azure Stack HCI deployment?

Question

Dear Community,

We've a hybrid Azure Stack HCI solution with spinning disks for capacity and NVMe disks for cache. Often we need to run heavy workloads for short periods, which quickly exhausts the cache.

The problem is that when we need to start a new workload the cache is already exhausted with data from old workloads, which we are not interested in and the new workloads are underperforming.

The Question:
How can we flush the disk cache stored on the NVMe disks, before adding a new workloads?

Answer 1

Hi @Ivaylo Petkov ,

There is no way to manually trigger an cache offload, except for retiring disks.

With an HDD+NVMe config the cache is flushed pretty quick to capacity. What makes you think the cache is full?
Did you check graphs in Windows Admin center?