Observability

WIP

Introduction

The ability to understand & measure the state of a system based on data generated by system. Helps to generate actions when unexpected scenarios happened.

How internal of system works? Otherwise it’s a black box Three core data pillars:

• Metrics Knowing the black box by asking why cpu spikees? why low memory? where throttle?..
  • MetricName: description of the name
  • Value value
  • Timestamp collected Time
  • Dimesion additional info
• Logs Records happened (past) and capture what event occurs
  • Timestamp
  • Message INFO:: XXX, DEBUG:: XXX, CRITICAL:: XXX
• Traces Where it’s travelled, how long it takes, easy to track it know hop by hop
  • TraceId: Unique
  • Spans: Indiviaul events forming a trace
    • startTime
    • parentId

Q) How client can rely on us? A) By telling quantifiable level of reliability of the service - 99.9% Availability:: permits a system to be offline for up to 8 hours, 45 minutes, and 57 seconds per year

SLO

SLA

SLI

https://github.com/ntkathole/banking-feature-store https://github.com/aniketpalu/feast-credit-score-local-tutorial/blob/main/demo_walkthrough.ipynb https://github.com/shap/shap