Let's Learn to Identify Technical Requirements for Better Design
summary
Claims 69% of IT projects fail largely from vague nonfunctional requirements. Two case studies show choices and costs: a kebab ordering app required “never crash” but only needs uptime during opening hours so Cloud Run + Cloud SQL meets needs at roughly 1/21 the cost of a Spanner/App Engine design, and an AWS e‑commerce build (Lambda, Kafka, API Gateway, S3, DynamoDB) scales but balloons in cost, later simplified toward a monolith. Practical fixes shown: measure 95th percentiles not averages, use timeouts circuit breakers retries for resiliency, compare SLA designs with cost estimates.
AI-written summary · about
chapters (8)
- 0:00Introduction and project failure rates
- 4:42Beyond functional requirements
- 11:12Availability requirements explained
- 19:19Case study cost and scalability issues
- 25:42Balancing business revenue and costs
- 30:20Designing for simplicity and evolution
- 37:51Observability and risk mitigation
- 43:29Key takeaways and challenges
related talks
36mHow to make technical decisions?
27m“Good Enough” Architecture Part 2
35mTell me why - About the purpose of architectural work
25mBuilding Applications in the Cloud
22mSimplicity: The Key to Software Success
59mMastering the System Design Methodology
46mDigital scale-up architectural patterns
44mThe Art of Architecting for Scale