The Art of Architecting for Scale
summary
Claim: modernizing code is not the same as splitting into services — distribution changes failures from logic bugs to availability problems. Recommendations: design for failure and scale (caching plus change‑data‑capture to keep payments working when account service is down), assign end-to-end ownership for customer journeys, drive architecture by measured SLAs and nines, enforce a non‑negotiable zero‑downtime migration path (strangler + CDC), and exercise systems with game days (chaos engineering). Example: ArgoCD + IaC rebuilds a cluster in ~25–30 minutes, revealed missing GitLab bootstrap.
AI-written summary · about
chapters (15)
- 0:00Architecting for scale intro
- 2:06Why systems feel slower
- 7:06Misunderstanding modernization and distributed systems
- 10:40Reason to adopt distributed
- 12:26Complicated vs complex
- 16:39Reliability vs availability
- 19:50Stable production illusion
- 23:02Failure and resilience
- 25:11Customer journeys ownership
- 28:43Platform teams SLAs
- 30:28Legacy migrations
- 32:55Zero downtime migrations
- 33:39Disaster recovery chaos
- 36:49Five key principles
- 38:34Q&A code duplication
related talks
01h 25mCreating an Incremental Architecture for your System
27mArchitecting for Change Debate
41mCultivating Architecture in an Agile World
42mSoftware Architecture in the Age of Agility and DevOps
38mAgile Architecture
49mCase study of a good enough architecture
01h 00mAgile Architecture
24mHow to design your team for scale