Completed
Real-world failure study
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Automatic Generation of Runtime Checkers for Production Distributed Systems
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Runtime checker (aka. detector/monitor)
- 3 Importance of runtime checker
- 4 Current checking practice
- 5 Complex internals of modern software
- 6 Common to exhibit gray failures
- 7 A real-world gray failure
- 8 Failure root cause
- 9 Ideal runtime checkers
- 10 A new approach
- 11 Panorama: capture in-situ observability
- 12 Convert a program into in-situ observer
- 13 Identify observation boundary and identities
- 14 Extract evidence
- 15 Example of analysis
- 16 Detecting real-world gray failures
- 17 Timeline of detecting failure case f1
- 18 Latency overhead to observers
- 19 Program reduction approach
- 20 Why doing reduction?
- 21 identify long-running regions
- 22 select checking target candidates
- 23 reduce long-running methods
- 24 encapsulate checkers
- 25 insert watchdog hooks
- 26 Prevent side effects
- 27 Watchdog generation
- 28 Failure detection evaluation setup
- 29 Detecting real-world failures
- 30 Silent semantic violations
- 31 Real-world failure study
- 32 Oathkeeper: detect silent semantic violation
- 33 How to express semantics?
- 34 Oathkeeper workflow
- 35 Emitting semantic event traces
- 36 General semantic rule templates
- 37 Extracted semantic rules
- 38 Runtime overhead
- 39 Conclusions
