Designing a Read-Intensive Architecture for Information Systems
summary
Claims concrete limits and fixes: Kubernetes microservices hit tail latency past ~1,000 nodes and a hostile example is 1,000 app ports × 100 cache nodes = 100,000 connections exhausting FDs and switches. Remedies include circuit breakers with 20 ms timeouts, per-pod in-memory projections with consistent hashing, zero-copy formats like flatbuffers and layout tuned for CPU, and a target hit rate where 85% is served from local cache and 1% touches the database. The design trades operational complexity and triple‑AZ overhead for dense hardware efficiency.
AI-written summary · about
chapters (11)
- 0:00Introduction to High Data Throughput Systems
- 0:14Understanding Modern Lead Intensive Architecture
- 1:18Challenges at Scale: Latency and System Issues
- 2:15Orchestration and Resource Management
- 3:36Critical Metrics and Performance Zones
- 4:25Architectural Solutions and Painkillers
- 6:19Optimizing Data Handling and Storage
- 7:45Advanced Data Optimization Techniques
- 9:53Shifting to High-Level Architecture
- 13:10Balancing Economics and Complexity
- 14:33Conclusion
related talks
59mDesign Trade-offs in Modern Architectures
41mDesign (Failures) Case Study
21mBuilding Distributed Platforms at Extreme Scale on Fast Changing Data
49mLarge-Scale Architecture: The Unreasonable Effectiveness of Simplicity
01h 00mTrading at light speed: designing low latency systems in C++
59mMastering the System Design Methodology
37mDeep Dive - Architecting for Million Operations Per Second!
41mBuilding the Future Trends in Modern Application Architecture