Zero-Downtime by Design: High-Availability Migration Patterns
summary
Claims a full legacy gateway migration to Kong completes with zero customer-facing incidents using an eight-phase Terraform-driven rollout. Validation uses a shadow traffic parity with parity-id and NoSQL store that samples roughly 1/5 of production requests to avoid doubling back-end load. Built interceptors to skip double execution for non-reads, automated difference detection for fixes, and reports 20–40% infrastructure/compute cost reduction while retaining compliance, vendor patching, and reduced meantime-to-recovery.
AI-written summary · about
chapters (16)
- 0:00Introduction and Speaker Background
- 0:11The Challenge of Zero Downtime Migration
- 0:37Case Study: Legacy Gateway Migration
- 1:04Defining Zero Tolerance for Downtime
- 1:26Risks and Solutions in Migration
- 2:27Legacy System Overview
- 4:38Migration Challenges and Validation
- 5:05Shadow Traffic Framework
- 8:10Eight-Phase Implementation Plan
- 10:30Shadow Traffic Findings
- 11:59Operational Impact and Benefits
- 14:03Broader Architecture Benefits
- 15:24Key Success Factors
- 16:22Lessons Learned
- 17:39Conclusion and Future Implications
- 18:34Q&A and Closing Remarks
related talks
40mStrategies for Zero Down Time, Frequent Deployments
30mIs Zero-Downtime Migration Possible? Moving From Ingress & Sidecars To Gateway API
50mZero Downtime Migrations with Spring Boot
37mZero-downtime Deployment Strategies (...) [ENG]
43mCrawl/walk/run: Migrating to the cloud
31mZero Downtime Migrations at Scale
31mHow We Went From Two Major Outages to 99.98% Reliability in 6 Months
35mMigrating a live site across the country without downtime