Plan overall CPU, memory and storage capacity needed to support projected system growth.
A server's overall capacity headroom is only as good as its most constrained resource — even if CPU, disk and network all have plenty of room, a server pinned at 95% memory utilization is effectively out of capacity. This calculator therefore scores overall capacity as 100 minus the highest of the four utilization percentages, and explicitly names that highest-utilization resource as the bottleneck, since that is the one that needs attention (or additional capacity) before any of the others.
Capacity score
score = 100 − max(cpu%, memory%, disk%, network%)
Averaging would hide a critical single-resource bottleneck behind three healthy metrics — a server can be functionally out of capacity even with low average utilization if just one resource (commonly memory or disk) is pinned near 100%, so the true constraint is always the highest individual figure.
Address that specific resource first — add RAM or tune memory-hungry services if memory is the bottleneck, expand or clean up storage if disk is the constraint, and so on — since improving non-bottleneck resources provides no additional real-world capacity until the bottleneck itself is resolved.
Use sustained averages over a representative window (an hour or a business day) rather than instantaneous snapshots, since brief spikes are normal and don't necessarily indicate a capacity problem — most monitoring systems track p95 or similar percentiles over rolling windows for exactly this reason.