Hyper Speed: When Big Data Blooms
summary
Report URI ingests browser telemetry and now handles spikes of almost 65,000 JSON payloads per second, routinely receiving one million POSTs in 11.2 seconds per million. Inbound reports are normalized, deduped and stored in Azure Table Storage with 90 day retention (3.9 TB). A global 8.4 GB bloom filter (k=7 hashes) held in Redis Bloom gives fast “definitely no / probably yes” checks, and sketches/top‑k/hyperloglog handle frequency and distinct counts in realtime.
AI-written summary · about
chapters (6)
- 0:00Browser security and big data
- 12:57Content Security Policy vs XSS
- 23:09Scaling CSP reporting
- 31:34How Bloom filters work
- 47:19Practical Bloom filter uses
- 51:32Q&A
related talks
01h 00mHyper Speed: When Big Data Blooms
56mWhat We’ve Learned From Billions of Security Reports
40mPerformance Testing and Big Data
08mReal-Time Data Systems: Speed, Scale, and Resilience
27mBatching Was Yesterday: Real-Time Tracking & Analysis For 100+ Million Visitors
09mBig Data, Big Performance
11mData science: Next-gen performance analytics
51mWhat We’ve Learned From Billions of Security Reports