Google-Scale Systems Lessons
Build practical intuition for large-scale software systems through Jeff Dean's 2010 Stanford talk on Google's search infrastructure, GFS, MapReduce, Bigtable-era design, reliability, latency, and performance tradeoffs.
3 sections ยท 9 lessons
Course outline
Search Scaling
- Read Growth Vectors
- Partition Query Work
- Move Hot Paths to Memory
- Plan a Search Scale-Up
Storage Primitives
- Make Failure Routine
- Hide Distributed Mess
- Shape Sparse State
- Choose Data Primitives
Reliability Patterns
- Software Owns Reliability
- Estimate Before Building
- Control Fanout Tails
- Defend an Overloaded Service