SLA (Service Level Agreement)
An SLA (Service Level Agreement) is a contractual guarantee of platform uptime and performance. If the provider fails to meet the SLA, the customer is entitled to service credits or other remedies defined in the contract.
SLA tiers in game server hosting
- 99.9% (three nines) — Up to 8.76 hours of downtime per year. Common for self-serve cloud tiers.
- 99.95% — Up to 4.38 hours per year. Typical for managed game server platforms.
- 99.99% (four nines) — Up to 52.6 minutes per year. Enterprise-grade, requires multi-provider failover.
- 99.999% (five nines) — Up to 5.26 minutes per year. Extremely rare in game server hosting.
What an SLA actually covers
Read the fine print. Some SLAs only cover the API control plane, not individual game sessions. Others exclude “scheduled maintenance windows” from the uptime calculation. The most meaningful SLAs cover end-to-end availability: the API responds, containers start, and players can connect.
How multi-provider architectures enable higher SLAs
A single-provider platform (e.g. running only on AWS) inherits that provider’s outage risk. If AWS us-east-1 goes down, so does your game. Multi-provider architectures distribute sessions across independent providers in the same region. If one provider fails, sessions route to healthy providers automatically. This is how platforms achieve 99.99% without relying on any single provider’s uptime guarantee.