https://github.com/sebastianhaeni/srecon-dublin-2023

Notes taken during SREcon Dublin 2023
https://github.com/sebastianhaeni/srecon-dublin-2023

Last synced: about 12 hours ago
JSON representation

Notes taken during SREcon Dublin 2023

• Host: GitHub
• URL: https://github.com/sebastianhaeni/srecon-dublin-2023
• Owner: sebastianhaeni
• Created: 2023-10-15T14:03:48.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-10-15T14:05:17.000Z (over 1 year ago)
• Last Synced: 2023-10-16T15:21:41.543Z (about 1 year ago)
• Homepage:
• Size: 5.86 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Notes on [SREcon EMEA Dublin 2023](https://www.usenix.org/conference/srecon23emea) 2023

## Key Takeaways

- MLOps: Focus on monitoring outputs and user impact.

- Monitoring/SLOs: Utilize histograms for better analysis.

- SLOs: Maintain a balance between reliability, on-call health, and productivity with SLOs.

- Automation as a Team Player:

  - Ensure automation processes are predictable and transparent.

  - Avoid creating surprises with automation and maintain simplicity.

- FinOps: Use tools like OpenCost/Kubecost to analyze and optimize cloud resource usage.

- Profiling: Utilize tools like Parca for continuous profiling and analysis.

## Favorite Quotes

> Looking at your average latency is like looking at the average temperature in the hospital.

> "Percentage availability" is for tourists.

> Instrument first, ask questions later.

> Try to build amplifiers and not prostheses.

## [Day 1](day1.md)

- Liz Rice's keynote highlighted the significance of eBPF technology.

- Symptom-based alerting for machine learning emphasized monitoring outputs, user impact, and input data distribution. Resources include a Google research paper and various metrics for monitoring ML systems.

- Reliability-enhancing procedures included steps such as service definition, SLOs, and operational response testing.

- Understanding QUIC vs HTTP/3 and its application mapping was also discussed, providing a GitHub resource for further exploration.

- Deploying and debugging HTTP/3 was briefly touched upon.

## [Day 2](day2.md)

- SRE principles applied to cybersecurity highlighted the importance of SLOs for security.

- The significance of distributed tracing and the importance of statistics for engineers were discussed. Resources included explanations on histograms and SLO measurement tools.

- Incidents as a means to enhance reliability investment were emphasized, focusing on mapping system capabilities and understanding user context.

## [Day 3](day3.md)

- Various talks covered topics such as OTel implementation, open-source observability, continuous profiling, Prometheus histograms, accountability engineering, and FinOps.

- The importance of automation being a good team player was stressed, citing the "Ironies of Automation" paper and the need to avoid creating surprises with automation.