Meridian Edge · Reliability
SLOs metReporting period Apr 1 – Jun 30, 2026 · Prepared by Reliability Engineering · Published Jul 3, 2026
The edge fleet met every top-line service objective through Q2, sustaining 99.962% availability across five regions while absorbing a 23% quarter-over-quarter increase in request volume. Two customer-visible incidents consumed a combined 0.9% of the quarterly error budget — well inside tolerance — but both traced to the same origin-failover path, which is now the top remediation priority for Q3. Tail latency in ap-southeast and sa-east remains the weakest surface and is the main reason the p99 objective closed the quarter in a warning band rather than green.
Attainment against the quarterly targets. Bars fill toward the target band; tone encodes headroom — green comfortably inside, amber approaching the line.
Per-region response-time percentiles and error rate for the reporting period. Latencies in milliseconds, measured at the edge.
| Region | p50 | p95 | p99 | Error rate | Requests |
|---|---|---|---|---|---|
| us-east | 18 | 74 | 198 | 0.011% | 4.82B |
| us-west | 21 | 88 | 224 | 0.014% | 3.11B |
| eu-central | 24 | 96 | 241 | 0.019% | 2.74B |
| ap-southeast | 31 | 129 | 318 | 0.042% | 1.36B |
| sa-east | 39 | 152 | 361 | 0.058% | 0.61B |
Committed edge capacity (Gbps) split across regions against the total provisioned pool. The gap between the total bar and the region stack is standing headroom.
Customer-visible events during the period, by severity. Two reached SEV-2 or higher; both routed through origin failover.