Domain 3 · Optimize Usage and Cost
Right-sized to actual demand, not Friday-afternoon guess.
Rightsizing matches resource size to real workload. CloudMonitor produces workload-aware recommendations across VMs, App Services, SQL, AKS, and storage — ranked by saving, routed to the owner, applied with one click instead of a forty-step runbook.
The problem
Recommendations that engineers ignore.
Generic recommendations.
"Downsize this VM to half" — based on a fortnight of CPU averages, ignoring the month-end spike that's the entire reason the VM exists.
Same advice prod and dev.
Dev gets the same conservative recommendation as production. The biggest wins — auto-shutdowns and aggressive sizing in non-prod — never appear.
Engineers see only noise.
A hundred low-confidence suggestions a week. The team mutes the email and the queue ages out untouched.
How CloudMonitor answers
Recommendations engineers will action.
Workload-aware sizing.
Recommendations across VMs, App Services, SQL, AKS node pools, and storage tiers. Each one weighs peak, percentile, and seasonality, not just the mean.
Saving-ranked queue.
Sorted by dollar impact, not alphabetical resource id. The top of the list is always worth opening.
One-click apply.
Approved recommendations execute via Azure APIs — no portal click-through, no ARM template hand-edit, full audit log.
Owner routing.
Each recommendation lands with the team that owns the workload, via the cost-group tree. No central queue, no triage tax.
Outcomes
Savings the team will actually take.
5
Resource families covered
1-click
Apply, audited
Owner
Routed via cost group
See workload-aware rightsizing in the demo tenant.
Walk the queue, approve a recommendation, see the audit log.