SLA
A commitment to customers about uptime and performance — if you miss it, they get refunds or credits.
Think of it like
A warranty on your service — 'we promise 99.9% uptime or your money back.'
Example
An SLA might be '99.5% uptime and p99 latency < 500ms' — you monitor and refund if you miss either.
How it actually works
SLAs are business contracts backed by engineering discipline. They drive infrastructure decisions (redundancy, monitoring, autoscaling). Exceeding SLAs is good; violating them costs money. SLA != SLO (objective) — SLA is the promise, SLO is what you aim for internally (usually higher).
For product teams
SLAs are your customers' recourse; enforce them internally with even stricter SLOs.
For engineers
Build infrastructure to meet SLAs reliably; monitor like your paycheck depends on it.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome