Tackling tech debt
summary
A concrete playbook: visualize tech debt with DORA metrics and tooling like Swarmer and LinearB, benchmark externally, secure sustained allocation of engineering time, and refactor incrementally rather than rewrite. Cycle time falls from 5.6 to ~2.5 days, throughput rises ~270%, and 80% of core development previously sits behind a monolith called damos. Training and an API‑first pull‑out strategy produce measurable latency drops and about half the monolith traffic. cycle time drops 40% no big rewrite
AI-written summary · about
chapters (18)
- 0:00Introduction
- 2:11Defining tech debt and its causes
- 5:42Legacy vs greenfield
- 7:06Dev experience impact
- 9:16Damos monolith case
- 11:26Risks of big rewrites
- 12:51Five-step approach
- 13:12Visualize and benchmark
- 15:20Selling tech debt
- 16:45Securing resources
- 17:07API-first strategy
- 18:15Board impact tracking
- 19:41Measurable improvements
- 20:42Daily debt practices
- 21:45Seniority and debt
- 23:08Incremental platforms
- 24:33Convincing others
- 24:55Summary
related talks
17mHow to deal with technical debt in fast-paced environment
26mTech debt as innovation, reframing this forever problem as an opportunity
20mPrioritizing Technical Debt as if Time and Money Matters
29mPrioritizing Technical Debt as If Time & Money Matters
40mFrom Strategy to Flow: A Journey Using DDD to Achieve Flow
51mCompleting the Rewrite from Hell: Five Years of Technical Debt and How We Escaped
32mScaling Developer Experience at Reddit, Airbnb and Amplitude
45mChanging attitudes toward legacy code