Disk I/O is the read and write activity between the operating system and its block devices. Agents typically expose cumulative byte counters per device, and the dashboard converts them into rates such as megabytes per second.
Throughput is only half the picture, and often the less interesting half. A device can be fully saturated at a low byte rate when the access pattern is small and random, because spinning disks and network attached volumes are limited by operations per second and queue latency rather than raw megabytes. Cloud volumes add a second trap: burst credit models let a volume run fast for hours and then fall back to a much lower baseline, so the same workload gets slow with nothing changing in the application. Read disk I/O next to iowait and load average instead of on its own.
Most of the alerting value here is in correlation and capacity planning. A sustained write rate sitting near a volume's provisioned ceiling predicts the incident days ahead of it. Hyperping collects read and write bytes for every block device, which makes per device rates and the noisy neighbor pattern of one volume dominating a host visible on the same timeline as CPU and memory.
