Distributed Systems
summary
Distributed systems fail for concrete, measurable reasons: clocks drift by 10 minutes across machines (DataDog), internal calls run about 100x faster than remote messages, and REST ~20x slower than RPC, so interfaces and message size decide performance. Queues trade speed for reliability, chaos engineering injects faults at low-traffic times to reveal race-driven bugs, observability replaces naïve monitoring, and diagnose by forming hypotheses and trying to disprove them rather than dumping every log.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 0:44Back in the 90s
- 2:36Moores Law
- 5:40Distributed Systems
- 6:48Problems
- 10:45Consistency
- 12:12Reliable vs Unreliable Messaging
- 15:39Message Transfer
- 17:48Errors
- 19:04Monitoring
- 21:07Narrative
- 27:24Coincidence
- 29:48Why Microservices
- 35:54Cloud Native Attitude
- 36:57Chaos Engineering
- 39:21Micro Services
- 41:17Supporting Multiple Languages
related talks
42mFive Common Mistakes With Distributed Systems
29mFollowing the Data: A Deeper Look into Distributed Applications
01h 02mConsistency and Agreements in Distributed Systems
01h 03mMicroservices vs The Distributed Monolith
42mArchitecting for Failures in Microservices
40mCAPtivating architecture: Navigating Distributed Systems and Microservices
01h 02mAsynchronous Distributed Systems
38mTechniques and Tools for a Coherent Discussion about Performance in Complex...