Managing a Billion Transactions with Tech, Tactics, and Teamwork
summary
Claim: over 1 billion requests across Black Friday 2023 with >99.99% availability. Architecture mixes VMs and Kubernetes, uses Google Pub/Sub, k6 operator for staged production load tests ramping to six times capacity, and GCP premium network tier to cut latency in Asia. Monthly production testing plus manual pre-weekend scaling prevents outages, reduces services from ~1,200 to ~400, and shrinks cloud spend (roughly €60k/month saved) — the setup works in practice.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:53Black Friday importance
- 5:01Closing data centers
- 7:09Twelve to fifteen factor apps
- 11:23Locate and Black Friday
- 12:06Black Friday prep
- 17:28Load testing in production
- 21:39Tech stack overview
- 26:33Past issues resolved
- 27:34Future scaling plans
- 30:02Premium networking Google Cloud
- 32:09Cloud cost reduction
- 33:54Key takeaways
- 36:04Q&A
related talks
30mBuilding Stable, Scalable & Agile Distributed Systems at Enterprise Scale
56mGetting all the 99.99(9) you always wanted
28mReliability Lessons from Running High Traffic Services for More Than 10 Years
54mGetting to 300 million requests a month with a one man AI/Dev-Ops team
31mHow We Went From Two Major Outages to 99.98% Reliability in 6 Months
36mHigh Availability at Braintree
26mHow We Prepped Our Infrastructure for a Super-Bowl Sized Traffic
43mTools and Team for 20 Billion PageView/Month