This topic describes performance metrics related to the time consumed by I/O operations on a host.
Background information
Host monitoring supports the monitoring of input/output operations per second (IOPS) on multiple mounted disks. If a host is mounted with multiple disks, multiple pairs of read and write metrics, such as sda: io_read and sda: io_write and sdb: io_read and sdb: io_write, are displayed. The metric field names and calculation expressions of the metric pairs are the same.
read
Definition
The average time consumed for each read.
Description
| Metric | Metric field name | Unit |
|---|---|---|
| read | io_read_time | Microseconds |
Calculation expression
1000000 * (avg(rate(node_disk_read_time_seconds_total{@LABELS}[@INTERVAL])) by (@GBLABELS)) / (avg(rate(node_disk_reads_completed_total{@LABELS}[@INTERVAL])) by (@GBLABELS))
write
Definition
The average time consumed for each write.
Description
| Metric | Metric field name | Unit |
|---|---|---|
| write | io_write_time | Microseconds |
Calculation expression
1000000 * (avg(rate(node_disk_write_time_seconds_total{@LABELS}[@INTERVAL])) by (@GBLABELS)) / (avg(rate(node_disk_writes_completed_total{@LABELS}[@INTERVAL])) by (@GBLABELS))