Do I have to migrate off my current APM to get these results?

No. We often start by layering OpenTelemetry + Prometheus alongside your existing APM. We instrument the critical path and route telemetry where it’s most actionable. You can phase out vendors on your timeline.

How long did this take end-to-end?

Four weeks to MVP: 2 weeks for instrumentation and collector rollout, 1 week for SLOs/alerts/dashboards, 1 week for runbooks and a game day. We scheduled the work so it didn’t block ongoing feature delivery.

Will this work if we’re not on Kubernetes?

Yes. The patterns (SLOs, burn-rate alerts, OTel traces, autoscaling on actual demand signals) apply on VMs, ECS, or serverless. The tooling changes slightly, the playbook doesn’t.

What about security and compliance for PII?

We scrub PII at the source using OTel processors, restrict egress from collectors, and keep payloads out of traces/logs. We’ve passed SOC 2 and PCI assessments with this setup at fintechs and healthtechs.

Case-studies · Nov 2, 2025 · 10 minute read

The Payroll Run That Didn’t Page Us: Observability That Stopped a Cascade Before It Started

How a pragmatic OpenTelemetry + Prometheus overhaul caught a Kafka-induced latency spiral 22 minutes before customers felt it—and kept the Friday payroll run green.

Back to all posts

The Payroll Run That Didn’t Page Us: Observability That Stopped a Cascade Before It Started

Key takeaways

Implementation checklist