8 min read · Published August 24, 2026 · Vast Networks
What belongs in a real SLA
An availability commitment with a stated measurement period. A repair or response objective, often expressed as mean time to repair. Performance objectives for latency, jitter, and packet loss where the service warrants them. A precise definition of what counts as an outage and when the clock starts. A list of exclusions. And a remedy with a process for claiming it.
If any of those are missing, the document is a description of intent rather than a commitment. The most commonly missing pieces are the claim process and the outage definition, and their absence is not accidental.
Availability percentages, converted to time
Percentages are designed to look similar. Converted into an annual downtime allowance they separate immediately: 99% permits about 3 days 15 hours; 99.5% about 1 day 19 hours; 99.9% about 8 hours 46 minutes; 99.99% about 52 minutes; and 99.999% about 5 minutes 15 seconds.
Then check the measurement period. The same percentage measured monthly and annually are materially different commitments — a monthly window resets the allowance twelve times a year, while an annual window lets one long outage consume everything.
Reading the exclusions, which is where the substance is
Scheduled maintenance is universally excluded and reasonably so; what matters is how much notice is required and whether there is a maintenance window you can plan around. Force majeure is standard. Customer premises, customer equipment, and customer-caused faults are excluded, which is fair — a router you power off is not a provider outage.
The clauses worth scrutinising are third-party access delays, "acts of third parties" drawn broadly enough to cover any fiber cut caused by someone else, and any exclusion for capacity constraints. A generous exclusions section can hollow out an impressive availability figure entirely.
What the remedy is actually worth
Remedies are typically service credits proportional to the outage duration — a percentage of the monthly charge, sometimes capped at one month. They are not compensation for consequential loss, and virtually every telecom agreement explicitly excludes consequential damages.
For an operation where an hour offline costs many times the monthly circuit charge, that is worth understanding plainly: the SLA provides accountability and a financial signal to the provider, not insurance. If the cost of downtime is high, the money is better spent on a diverse second path than on negotiating a stronger credit schedule.
Questions worth asking any provider
How is an outage detected — by your monitoring or by my call? What is the target response time, and is it a target or a commitment? Is maintenance excluded, and how much notice do I get? Are credits applied automatically or must I claim them, and within what window? What is specifically excluded that I should design around?
The answers separate providers far more reliably than the headline number. A provider that answers all six clearly is telling you something about how it operates.