Skip to content

SLA Comparison

A 99.99% uptime number on a marketing page is not the same as a contract with enforceable refund terms. This page grades 15+ developer infrastructure providers on three things: the stated uptime SLA, how refunds actually work when the SLA is breached, and how transparently incidents are reported.

Methodology

  • Stated uptime — the percentage value the vendor commits to in its public SLA document. Where SLA varies per service (AWS, Twilio), the headline number is the most-used product. null means no formal SLA exists.
  • Refund mechanism — automatic means credits are issued without a claim filed. Request means the customer must file within a window (usually 30 days). Credits-onlymeans the "refund" is a credit on future invoices, not actual money returned. None means there is no refund mechanism documented.
  • Public incident history (years) — how far back the status page and post-mortem archive go. Long histories indicate consistent operational discipline; short histories indicate either a young provider or one that periodically purges incident records.

Why the grade is not just the uptime number

A vendor stating 99.999% with a credits-only refund mechanism and a no-claim refund policy is, in practice, weaker than a vendor at 99.95% with automatic credits and a 10-year transparent status page. The grade is integrated: stated number, ease of collecting on the SLA, and historical credibility together.

What this score doesn't measure

  • Actual realized uptime — we report the stated SLA, not measured. For measured uptime see /outages.
  • Latency or performance SLAs. Uptime says "the API responds", not "the API responds fast."
  • Data durability — that's a separate SLA (typically 99.999999999% for object storage).

Reading the table

The Stated Uptime column shows the headline number. The Refund column tells you what you actually get when that number is missed. The Years column is the depth of incident history available. Notes explain anything that doesn't fit the numbers — tier exclusions, specific incidents, claim windows.

Rankings