Calculate network bandwidth utilization percentage against interface link capacity.
Bandwidth utilization expresses observed throughput as a fraction of the link's total rated capacity, giving a quick saturation signal independent of the absolute traffic volume — 420 Mbps means very different things on a 1 Gbps link (42%, comfortable) versus a 500 Mbps link (84%, elevated). Sustained utilization above roughly 70-80% on shared links typically starts introducing queuing delay and jitter even before hard packet loss occurs, since bursty traffic patterns mean average utilization understates peak instantaneous demand.
Utilization percentage
util% = actual_bps / link_capacity_bps × 100
`sar -n DEV 1` or `ifstat` show live per-interface rx/tx rates; `cat /proc/net/dev` gives cumulative byte counters you can sample twice and divide by the elapsed time for an average rate over that interval.
Raw throughput without context doesn't reveal whether a link is comfortably provisioned or about to become a bottleneck — utilization normalizes for link speed, making it directly comparable across links of different capacities and a better trigger for capacity-planning alerts.
Brief spikes to 100% during legitimate bursts are normal and not necessarily harmful, but sustained 100% utilization means the link is the bottleneck for all traffic crossing it, causing queuing delay, increased jitter, and eventually tail-drop packet loss once buffers fill.