Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Scaling Distributed Tracing

Scaling Distributed Tracing

A talk on scaling tracing at the distributed tracing workshop in Seoul - July 2019

5648ef74e40ec5c26fd0b33f2f961038?s=128

Autoletics

July 11, 2019
Tweet

Transcript

  1. WILLIAM LOUTH PRODUCT R&D DISTRIBUTED TRACING WORKSHOP SEOUL, SOUTH KOREA

    JULY 2019 Scaling Distributed Tracing
  2. Company

  3. Product

  4. Product

  5. @Autoletics

  6. Technologies

  7. Research Controllability Resilience QoS for Apps Resource Management Self-Adaptive Observability

    Service Signals Episodic Memories Operability Visualizations Learning Intelligence Microservices Containers Event Architectures Reactive Systems Serverless Actors
  8. None
  9. Observability Distributed Tracing Metrics Logging Reality Activity Metering Mirrored Simulation

    Branch Signalling Behavioral Signalling Model Reconstruction • Explorative • Developer • Machine Operational • Effectiveness • DevOps • Human
  10. Measurement Model Memory Observability

  11. Scaling M easurem ent Overhead M odel Transport M em

    ory Storage Accuracy Attention Significance
  12. Scales Short Long Deep Shallow Small Big Local Remote Machine

    Human Signal Data
  13. Cybernetics Observability Controllability attention + action perception

  14. Coordination Observability Controllability Monitoring Management Strategic Tactical

  15. Cognition Information Significance Observability Monitoring Sensory Sem antics

  16. Communication Objects Observer Signals State Inference

  17. Collection Memory Model Measurement

  18. Collective M em ory M odel Measurement M3 M3 M3

    M3 M3 M3
  19. Conduction Instrument Measure Collect Transmit Store Data • Cost •

    Time
  20. A B C D E Trace

  21. Transmit A B C D E M3

  22. Federate A B C D E M3 M3 M3 M3

    M3
  23. Backtrace A B C D E M3

  24. None
  25. None
  26. Sample Random Conditional Windowing

  27. Limit 1 2 3 4 12 34 1 2 3

    4
  28. None
  29. Discriminate

  30. Buffering

  31. Drop

  32. Discard

  33. Delay

  34. Degrade A B C D E A B C A

  35. None
  36. Costs Clock Tag Stack Log Item Id