Web-Scale IT Engineering.
summary
Argues enterprise IT shifts to Google-style web-scale architecture: replace specialised appliances with commodity servers, software-defined everything, and local data replicas so each server keeps hot copies and the system self-heals. Claims typical vendor clusters cap at 64 servers while top-100 adoption hits ~70%. Describes rewriting Cassandra over ~three years into a distributed file system, doubling RocksDB-style performance, and delivering multi‑million IOPS in VMs to prove the approach at scale.
AI-written summary · about
chapters (17)
- 0:00Introduction and background
- 1:55Enterprise IT silent revolution
- 3:44Why Yahoo lost to Google
- 6:15Traditional IT limitations
- 8:20Web-scale IT adoption growth
- 11:59Web-scale IT origins and principles
- 17:08Hyper-converged infrastructure explained
- 20:25Hardware reliability and software redundancy
- 23:18Nutanix data fabric architecture
- 26:52Virtualization vs containerization
- 31:08NoSQL and distributed file systems
- 34:41Optimizing KVM for performance
- 36:46Million IOPS in virtual machines
- 38:12Software upgrades and performance
- 40:22Containerization and Kubernetes
- 43:16DevOps stack and orchestration
- 45:09Mission-critical applications
related talks
29mWhen Web Scale Doesn't Make Sense
38mThe Dark Side of Scale
01h 05mSpeed & Scale: How to Get There
49mHypergrowth scaling made simple
42mLarge-Scale Architecture: The Unreasonable Effectiveness of Simplicity
38mWhat Comes After web-scale?
56mMastering Cloud Scalability and System Designs Deep dive
20mSoftware Performance Optimization and Scaling