Keynote: Federated Analytics at Scale
summary
Claim: a prototype called Worldwide Herd runs federated analytics by pushing code to gateways and data zones, computes local histograms (then merges counts) to reduce bandwidth and preserve privacy, and records every job and data hash in a blockchain ledger for trust and reproducibility. It integrates with Ajax Foundry and vmware realize platform, supports containers anywhere, and enforces a minimum-site sample cutoff (no sites with ≤5 samples). Expected benefit is near real‑time, global analytics without centralizing terabytes or the projected one trillion IoT endpoints.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:54Artificial intelligence evolution
- 4:17Learning algorithms
- 4:29Common denominator
- 7:23Edge-core/fog-cloud continuum
- 9:15Data distributed and stranded
- 14:15Federated analytics main challenges
- 17:47Federated computing framework
- 17:59Federated analytics and learning
- 18:24World Wide Herd (WWH) unique approach
- 24:17Virtual data and analytics fabric
- 27:35Inspired on Greek methodology
- 27:46Inspired on WWW principles
related talks
38mAdd ALL the things: abstract algebra meets analytics
27mStream, Store, Visualize: Fast-Track Real-Time Analytics the OSS Way
27mTales at Scale: Analytics at 1000 QPS and Beyond
19mInsights for Everyone - Democratising Analytics
37mPanel: The Hidden Value behind Cryptoanalytics
31mFeast as a Cloud Native Data System: Bridging Batch Processing and Online Serving...
05mThe Future of Data Warehousing
55mKeynote – Joseph Sirosh