Betting On Evolutionary Architecture
summary
Claims that placing small bets with continuous delivery and infrastructure-as-code beats one big upfront architecture decision; built an S3+XSLT pipeline, extracted an S3 + XSLT transform microservice, and iteratively tests options with Vegeta/ab and metrics dashboards. Starting page loads ~35s, increasing CPUs, moving compute to the data and tuning drops to ~3.5s, adding a Varnish cache in front of the service hits ~0.2s but cache invalidation forces rolling back to a cache-in-front-of-S3 compromise (~150ms). Continuous deployment and observability make these experiments cheap and reversible.
AI-written summary · about
chapters (10)
- 0:00Start
- 14:35how do you decide which path to follow?
- 19:15lets just add a cache
- 24:36at do we mean when we say "monitoring"?
- 31:44Cloud native infrastructure as code
- 32:51Putting it all together
- 34:26Second Guess: Moving compute to the data
- 35:16ird Guess: Transformations are slow and could be optimised
- 37:13Cache in front of content service
- 40:05Options, Small Bets
related talks
52mBetting on Evolutionary Architecture
59mBetting on evolutionary architecture
49mCase study of a good enough architecture
33mBuilding Evolutionary Architectures
50mTowards an Evolutionary Architecture
28mThe lean architect: Experimenting your way to better decisions
39mPrinciples of Evolutionary Architecture
26mArchitecting for scale