How Is My Disk Behaving?
Throughput — Read / Write
Operations (IOPS) — Read / Write
Disk Utilization
API Reference
Console Panel
metricType
Last updated
All disk panels are scoped per attached volume — pick the volume from the dropdown at the top of the section.
Bytes per second that the guest is asking the storage layer for, measured on the host side of the block device. This is the ground-truth view, independent of anything the guest reports.
Read and write requests per second. Read together with Throughput to compute the average request size — small-block random I/O and large-block sequential I/O behave very differently and should be reasoned about separately.
How busy the block device is, derived from cumulative service-time counters. This is the panel that distinguishes "my disk is slow" from "my disk is busy":
High utilization at high IOPS — the workload is saturating its disk tier. Reduce the request rate, batch writes, or upgrade the disk tier.
High utilization at low IOPS — request rate is modest but each request is slow. This points at the storage backend; contact support.
All disk metrics use DescribeDiskMonitorData with the metricType values below.
Throughput — Read
DISK_READ_BYTES
Throughput — Write
DISK_WRITE_BYTES
Operations (IOPS) — Read
DISK_READ_OPS
Operations (IOPS) — Write
DISK_WRITE_OPS
Disk Utilization
DISK_UTIL
Last updated