📖Designing Data-Intensive Applications
–
- authors
- Martin Kleppmann
- year
- 2017
- url
- https://www.amazon.com/dp/B06XPJML5D/
p.28 In fault-tolerant systems, it makes sense to increase fault rate (deliberately introduce more faults, kill servers, etc.), so that fault-handling code is executed and tested. Netflix’s chaos monkey is an example
p.38 latency vs. response time. response time = time between user issuing a request and receiving results (network delays, processing time (service time), queuing delay). latency is the duration that a request is waiting to be handled (during which it is latent)
response time = network latency + queuing latency + processing time
p.187 language-specific de/serialization frameworks often require ability to instantiate arbitrary classes which often leads to security vulnerabilities
p.212 data outlives code → Data outlives code