Synthesizing Sanity with, and in Spite of, Synthetic Monitoring
summary
Four years of in-house work yields a concrete stack: Pollinator (browser runners) and PM2 for orchestration, used by ~30–40 teams across Jira. Tests run on a subset of 300 shards after cost growth forces reduced granularity and frequencies (every 5–15m), because synthetic checks once generated over 10% of Jira requests. Best practices: avoid sleep, reuse login and helper libraries, declare check dependencies, soak in staging and fail CI, alert by aggregated thresholds. A synthetic run even reveals and helps fix an N+1, cutting load ~90%.
AI-written summary · about
chapters (14)
- 0:00Synthetic monitoring intro
- 1:06Speaker background
- 3:16What is synthetic monitoring
- 5:46Types and tools
- 10:20Testing pyramid and synthetic tests
- 15:58Why synthetic checks flake
- 20:11Synthetic vs real user monitoring
- 23:19Jira’s synthetic setup
- 27:52Granularity vs cost
- 31:45Post-deployment verification
- 33:47Alerting strategies
- 35:11Ownership challenges
- 37:58Improving synthetic monitoring
- 40:47Future of synthetic monitoring and AI
related talks
44mUnleash Synthetic Monitoring with Supercharged Test Automations
30mSynthetic Monitoring and E2E Testing: 2 Sides of the Same Coin
32mSynthetic Monitoring and E2E Testing: 2 Sides of the Same Coin
30mSynthetic Monitoring: Automating the User Experience for Continuous Improvement
26mSynthetic Monitoring and E2E Testing: 2 Sides of the Same Coin
21mChallenges, Best Practices, and Solutions for Monitoring and Alerting...
40mDecoding Synthetic Monitoring: A Journey from E2E UI Tests to Service-Level Probes
24mSmoke to avoid Fire