February 19, 2016 Book Reviews
The book has four parts – stability, capacity, general design issues, and operations. Each part is introduced by interesting case studies, for instance describing how single unhandled exception costed an airline a day of nightmare and tons of money.
Keywords
Let me list some of the keywords from the book for better idea what the book is about:
- Part I – Stability: scaling strategies; fail fast; SLA; timeouts; circuit breaker; bulkhead (don’t sink entire ship); common sense on log files; health checks; etc.
- Part II – Capacity: resource pools; threads; payload sizes; db performance; cookies; caching; etc.
- Part III – General Design Issues: load balancing; clustering; configurations; networking; security; etc.
- Part IV – Operations: transparency; statistics; logging; monitoring systems; etc.
Pros
Things that I really liked about the book:
- It easy to read, provided you have relevant experience
- Real world case studies
- “Remember This” sections after most of the chapters (better than boring summaries)
- Realistic strategies and approaches
- Business costs and impact discussed
There is a really nice strategy explained on zero-downtime. One, that especially boggles me, was approach related to database upgrade process for zero downtime. The strategy is relatively simple: 1) you extend the system/db and add some bits and pieces so it is backwards compatible, for db it could mean new nullable columns and triggers to fill-in data, 2) you roll out, 3) you cleanup. Easy!
Cons
Not so great things about the book:
- Some technologies mentioned are somewhat outdated (book released in 2007)
- Examples are in Java and frameworks mentioned are Java-world, so not totally language agnostic
- At places a bit too ops heavy
Conclusion
The book will be most useful for developers of large enterprise applications, especially where downtime costs real money.
But, in any case, I would highly recommend any developer, and deployment engineer for that matter, to read this book!
codemore code
~~~~