Completed
Increasing the size of my thought leadership
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Distributed Systems in Production: Tactics and Strategy - Lecture 32
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Distributed Systems in Production Jeff Hodges 2014-04
- 3 Why you should listen to me
- 4 Why you shouldn't listen to me
- 5 Scale-invariant
- 6 Building and running Distributed Systems
- 7 Quick foundation
- 8 What Makes Distributed Systems Different
- 9 Garbage collection spiral on a single machine causes requests to timeout • A process is overloaded, so too many clients get stuck trying to connect to it, so it gets slower • Socket write succeeds lo…
- 10 Partial Failure
- 11 "It's slow" is the hardest problem you'll ever debug
- 12 Metrics are the only way to get your job done.
- 13 On profiling
- 14 Deploys should change a metric
- 15 Logs are liars
- 16 Avoid coordination
- 17 If your problem fits in memory, it's probably trivial
- 18 Back-pressure
- 19 Dropping new messages on the floor • Returning documented overload errors until the system clears • Timeouts and exponential back-offs
- 20 Create partial availability
- 21 Search
- 22 Who to Follow in the monorail
- 23 Consider a private messaging database
- 24 Separating deploy from release
- 25 Roll out infrastructure with feature flags
- 26 Slow, dark rollouts
- 27 Multiple versions are the norm
- 28 Exploit data-locality
- 29 Extract services
- 30 Stricter boundaries means even less cheating
- 31 Pulling out a service makes deploys easier
- 32 Avoids human coordination costs that libraries require.
- 33 SOA through standardization
- 34 On-call rotations
- 35 The Notorious E.O.C.
- 36 Increasing the size of my thought leadership
- 37 Robust distributed systems cost more than undistributed systems.
- 38 Robust open source distributed systems are less common
- 39 Collaboration is politics
