Rapid Experimentation Simplified with Linkerd
summary
A financial‑services cloud team uses Linkerd traffic‑splitting plus CRDs to run simultaneous A/B/candidate experiments across versions, feeding metrics into observability via Prometheus/Grafana so regressions show up immediately; a live demo injects latency and produces a 10s P99 on v3, proving the split and dashboards work with no code changes. The setup supports canaries, chaos and federated meshes, and scales to hundreds of teams by lowering the developer DX barrier to run low‑cost experiments.
AI-written summary · about
chapters (7)
- 0:00Start
- 0:55Agenda
- 1:35Why Is There a Need for Experimentation
- 6:31Infrastructure Architecture
- 9:09Traffic Splitting
- 10:12Demo
- 11:09The Traffic Split
related talks
36mService Mesh: The Missing Link in Developer Workflow Efficiency
01h 15mExperimentation at scale: Fuel your application’s growth through A/B testing
26mML Product Experiments at Scale
31mBuilding a Data Experimentation Platform Without a Platform Team
33mA/B testing: performative experimentation at scale
50mAdvanced Pipelines for Hypothesis-Driven Development Smart Routers...
25mDevOps Handbook Experiments in Accelerating Delivery
49mHow to get smarter while doing your job