Description

Qemu to provide a configurable threshold (with a sane default) to raise QMP events and potentially pause the VM if the latency of a disk is surpassing the configured latency threshold.

Purposes:
1. Via these notifications, we want user to be made aware of latency issues in their storage back-end, so they can take measures to improve the situation.
2. Pausing the VM in singular cases of extreme latency, until those specific requests have completed on the host, may avert critical errors in the guest, like BSODs or kernel panics due to timeouts on critical I/O (e.g. swapping).

What SSTs and Layered Product teams should review this?