Build it fast, make it quick
summary
Warning against premature optimization with concrete counters: a legacy payment job grows from a few requests to 40,000 checks/hour, 15 minutes each and becomes unfixable; two sprints spent on UI tweaks allegedly costs $23M lost revenue while payment failures go unaddressed. Advice is to pick monoliths until load demands monolith first, microservices later, gather metrics (Google Analytics, logging, k6), define SLAs, prioritize business-impacting fixes, rotate one dev per sprint on tech debt, and measure cost versus benefit before rearchitecting.
AI-written summary · about
chapters (6)
- 0:00Start
- 11:04Optimization can be problematic
- 27:28Underlying Phenomena
- 32:06Improper Refinement
- 36:37What are the benefits?
- 49:48NEVER stop refactoring
related talks
46mDigital scale-up architectural patterns
26mArchitecting for scale
43mBeyond micro-services, embrace composable architectures
26mTackling tech debt
16mImprove application and code quality without overengineering
42mBetting On Evolutionary Architecture
48mMinimum Viable Architecture
24mExpediency & Fluency