Telecommunications
Five thousand authorisation decisions a second
A highly concurrent policy server with strict reliability guarantees, plus the telemetry that cut global operational support workload by 60%.
- 5,000/s
- remote-access requests, sustained
- 60%
- reduction in global operational support workload
- Federal
- reliability and auditability requirements
Remote-access authorisation for US federal clients: every request evaluated against a live rule set, with reliability guarantees that make “mostly correct” a failing grade.
Throughput without cutting corners
The server sustains 5,000 requests per second while evaluating a full policy rule set per request. The design work was in the parts that usually get skipped — rule evaluation kept off the hot path where it could be, contention removed rather than tuned around, and a concurrency model that degrades predictably instead of collapsing at the edge.
Observability as a first-class feature
The second half of the engagement was proactive monitoring telemetry integrated directly into the operational tooling, with automated alerting on the conditions that actually preceded incidents.
Global operational support workload fell 60%. Not because there were fewer problems, but because the system stopped requiring a human to notice them.