The Monitoring Cost Budget Discipline

Most teams spend 5-15% of infrastructure cost on monitoring. The budget discipline that catches monitoring sprawl before it doubles.

Healthy ratio

Monitoring spend as a percentage of infrastructure spend has a healthy band. 5-15 percent; under-instrumented below; over-instrumented above.

5-15 percent of infra spend. Per-org healthy band; below means under-instrumented, above means over-instrumented.
Track monthly. Per-month spend ratio; trend lines matter more than the absolute number.
Industry comparison. Per-industry benchmark: SaaS around 8 percent, high-availability closer to 15 percent; anchor against peers.
Per-quarter published ratio. Per-quarter visible chart; supports honest finance conversations rather than annual surprises.

Three cost drivers dominate. High-cardinality metrics, log volume, trace sampling.

High-cardinality metrics. Per-metric the single-biggest-cost-driver effect; audit cardinality quarterly.
Log volume. Per-service verbose-service tax; tune log levels in production rather than shipping debug traffic.
Trace sampling. Per-service 100-percent-rarely-needed rule; tune sampling to fit the budget.
Per-driver named owner. Per-driver responsible engineer; supports targeted optimisation rather than committee.

The budget policy makes spend visible. Monthly finance review, per-metric cost owner, quarterly retirement.

Monthly finance review. Per-month spend-with-finance review; same scrutiny as compute spend.
Per-metric cost owner. Per-metric named owner; owner is responsible for retiring if value drops.
Quarterly metric retirement. Per-quarter lowest-value metric retirement; budget rebalances against new priorities.
Per-policy published budget. Per-quarter visible monitoring budget; supports cost conversations with stakeholders.