Live Q&A Session – Anca Agape
summary
Q&A lists concrete practices and hard lessons: use the utilization saturation errors metrics to get cost‑per‑request; prefer queues for command buffering and streams for event logs with checkpointing as stored offsets; async platform roadmap moves to composable serverless building blocks and an POSIX API so Python, Go or Rust runtimes can be bundled; chaos experiments surface AZ imbalances, single‑replica agents, networking breakage, cold starts and CPU saturation causing scaling surprises; biggest work is integration, onboarding programs and prioritizing high ROI efficiency fixes.
AI-written summary · about
chapters (15)
- 0:00Q&A
- 0:26Speakers introduce themselves
- 3:18USE method for resource metrics
- 5:05Async computing platform effectiveness
- 6:50Future challenges for async platform
- 8:15Network workflow language support
- 9:17Network workflow platform specificity
- 10:41Chaos experiments reliability issues
- 13:30Chaos engineering platform challenges
- 15:53Meta cloud challenges
- 18:04Checkpointing in stream processing
- 19:30Network resource allocation monitoring
- 20:54Infra efficiency best practices
- 25:13Event versus command distinction
- 26:57Closing remarks
