SLA math: what 99.5% uptime actually means

The basic conversion

For a 30-day month (~43,200 minutes):

Leap years and exact day counts change the decimals; they do not change the story. Moving from 99.5% to 99.9% is not a “small bump.” It is often a five-fold reduction in allowed outage.

What customers actually feel

Customers rarely experience “percentage.” They experience:

• Can the branch take payments?
• Can the clinic sync records?
• Can the mine report telemetry?

A single 90-minute outage in peak hours can “fit” inside a 99.5% month and still destroy trust. SLA math without when the downtime lands is incomplete.

SLA vs. what the network delivers

Service agreements and path quality are related but not identical:

• SLA is a commercial promise (credits, remedies, measurement window)
• QoS / path health is what routers, underlays, and last-mile links actually deliver

You can meet a loosely measured SLA while users suffer jittery video. You can also over-deliver on uptime and still miss application SLOs. Design monitoring around user journeys, not only circuit “up.”

A simple decision framework

When choosing between tiers (or between primary fiber and satellite backup):

1. Write the allowed downtime in hours, not percent
2. Ask which hours are unacceptable (payroll day, clinic morning, shift change)
3. Price the engineering required to cut those hours—not the brochure percentage
4. Decide what you will measure (probe from where? synthetic? customer ticket?)

Honest sales language

Better than “we’re five nines”:

“Our commitment is 99.5% monthly on the primary path. Backup is designed for continuity, not identical throughput. Here’s how we measure both.”

That sentence prevents three future arguments.