Black Friday Brilliance: Managing a Billion Transactions with Tech, Tactics
summary
Loqate handles over 1 billion BFCM API requests with >99.99% uptime by combining Kubernetes and VMs, manual pre-scaling, and a caching layer; load tests run in production with chaos-monkey-style scenarios and a dataset spanning US/EU/Asia. Synchronous DB calls get buffered via Google Pub/Sub, analytics use BigQuery, and traffic routes across three data centers with Google’s premium networking tier as a stopgap. A prior 1,200-service sprawl falls to ~400, startup time improvements cut cloud spend, and careful telemetry drives per-service scaling.
AI-written summary · about
chapters (10)
- 0:00Introduction and speaker background
- 1:50Black Friday history and impact
- 5:06Company background and cloud transition
- 13:58Pre-Black Friday preparations
- 17:52Current tech stack and architecture
- 24:53Past issues and their solutions
- 26:43Future tech plans and cost savings
- 31:53Premium tier networking benefits
- 33:15Key takeaways and recommendations
- 45:44Q&A session
related talks
42mBlack Friday Brilliance: Managing a Billion Transactions with Tech, Tactics, and Teamwork
22mBlack Friday Brilliance: Managing a Billion Transactions with Tech, and Teamwork
33mHigh availability by offloading work
47mApplying “web scale” patterns (...)
40mFinOps Examples using Google Cloud BigQuery by Aliz.ai
56mGetting all the 99.99(9) you always wanted
24mBuilding Planet-Scale Payment Platforms
41mBuilding and running applications at scale in