DevOps Indonesia #14 - Monitoring Principles
1 like294 views
Monitoring Principles This document summarizes key monitoring principles, including anti-patterns to avoid and design patterns to employ. Anti-patterns discussed include tool obsession, treating monitoring as a separate job, checkbox monitoring without understanding what indicates health, and relying too heavily on monitoring as a crutch rather than fixing underlying issues. Design patterns promoted are having a composable monitoring platform using multiple specialized tools, starting monitoring from the user perspective, leveraging monitoring tools from vendors rather than building custom solutions, and viewing monitoring as an ongoing process of continual improvement. The presentation concludes with time for questions.
DevOps Indonesia #14 - Monitoring Principles
- 1. PAGE1 DEVOPS INDONESIA PAGE 1 DEVOPS INDONESIA DEVOPS INDONESIA DevOps Community in Indonesia Jakarta, 28 Januari 2019 Monitoring Principles
- 2. Subjects ● Monitoring Anti-Patterns ● Monitoring Design Patterns
- 3. Introduction ● William Notowidagdo ● Head of Engineering at Ralali.com ● Past companies: Amazon Web Services, HarukaEdu, ...
- 4. An anti-pattern is something that looks like a good idea, but which backfires badly when applied. —Jim Coplien
- 5. Anti Pattern #1: Tool Obsession ● There are no silver bullets ● Sometimes, you really do have to build it ● Avoid Cargo-Culting tools Image taken from https://refactoring.guru/design-patterns/prototype
- 6. Anti-Pattern #2: Monitoring-as-a-job ● Monitoring is not a job -- it's a skill ● And it's a skill everyone on your team should have
- 7. Anti-Pattern #3: Checkbox Monitoring When you have monitoring systems for the sake of saying you have them Common signs: ● The service still goes down without your knowing why ● You find yourself constantly ignoring alerts ● You are checking systems for metrics every 5 minutes
- 8. Anti-Pattern #3: Checkbox Monitoring (cont.) To fix this anti-pattern ● Monitor what does "working" actually mean ● OS metrics aren't very useful -- for alerting ● Collect your metrics more often
- 9. Anti-Pattern #4: Using Monitoring as a Crutch ● You're constantly adding more monitoring to a service ● Don't forget the next step: fixing the problems Image taken from https://vectorstock.com/15312737
- 10. Anti-Pattern #5: Manual Configuration ● "Can you add this to monitoring?" ● Services should self-register ● You should be able to quickly configure new checks/nodes
- 11. Design patterns are typical solutions to common problems in software design monitoring.
- 12. Pattern #1: Composable Monitoring Use multiple specialized tools and couple them loosely together, forming a monitoring "platform"
- 13. Pattern #2: Start from the User Perspective ● Start monitoring as close to the user as possible ● "How will these metrics show me the user impact?"
- 14. Pattern #3: Buy, Not Build ● You're (probably) not an expert at architecting these tools ● SaaS allows you focus on the Company's product
- 15. Pattern #4: Continual Improvement World-class isn't achieved in a week
- 16. Thank you Now it's time for Q&A