Instrument
The Cluster KPI.
A cluster is not the average of its VMs. Score each signal - the worst VM for depth, the cluster average for breadth - and the guide's progressive weightage combines them: a single red weighs as much as eight greens.
The signals
Each carries its own four-band score (100 = no contention, 0 = at your SLA-breach threshold). Depth reads the worst VM; breadth reads the cluster average - the guide deliberately scores the average, not the % of VMs.
CPU ready · worst VMdepth
12
the deepest CPU queue any one VM suffers · weight ×8
CPU ready · cluster averagebreadth
88
how widespread CPU queuing is · weight ×1
CPU co-stop · worst VMdepth
70
multi-vCPU scheduling pain, at its worst · weight ×2
CPU co-stop · cluster averagebreadth
92
co-stop across the population · weight ×1
RAM contention · worst VMdepth
85
the worst memory latency any VM sees · weight ×1
RAM contention · cluster averagebreadth
95
memory pressure across the population · weight ×1
Ballooned / swapped / zippedsecondary
90
sticky reclamation signals; balloon gets 4x the threshold of swap+zip · weight ×1
Dropped packets · worst ESXidepth
90
host-level, not per-VM; green is deliberately not zero · weight ×1
VMs being vMotioned (%)leading
88
the cluster shuffling VMs to cope is a leading indicator · weight ×1
Error packets · worst ESXioverride
100
rare but severe, so it sits in a second table that can only pull the KPI down, never lift it
The KPI
The progressively weighted, normalised combination - the number the guide would put on the cluster.
51 /100
yellow
naive average
79
what a plain mean would claim
the guide’s engine
51
red is heard 8x louder than green
What drags it most
CPU ready · worst VM12
CPU co-stop · worst VM70
RAM contention · worst VM85
Depth reads the worst VM because averages arrive too late; breadth reads the cluster average because the population matters. Signals use the 20-second peaks so short bursts are not missed. Bands: green 75+, yellow 50+, orange 25+, red below 25.
Cluster Performance (%): the KPI, its inputs and exclusions →Calculation: the four bands and progressive weightage →Aggregation: depth vs breadth →VM Performance (%): the 20-second peak metrics →
Official Broadcom docs
↗Cluster Compute Resource metrics (the full cluster metric reference)↗Virtual Machine metrics, incl. the 20-second peak definitions↗Configuring super metrics (the KPI implementation vehicle)↗Create a super metric (and enable it in a policy)↗Super metric functions and operators (max, avg, count)VCF Operations 9.1 documentation, each page verified. The guide notes the exact per-metric cluster thresholds are shown in its screenshots; the engine here (bands, weighting, override) is what the text specifies.