Episode

Continuous Feedback With Roni Dover

Podcast: Arrested DevOps
Published: Jul 28, 2022
Duration seconds: 3064
Processing state: processed
Canonical source: https://www.arresteddevops.com/continuous-feedback/
Audio: https://media.blubrry.com/arresteddevops/content.blubrry.com/arresteddevops/arrested-devops-podcast-episode183.mp3
JSON: /v1/public/podcasts/arrested-devops/episodes/continuous-feedback-with-roni-dover
Markdown: /podcast/arrested-devops/continuous-feedback-with-roni-dover.md

Actions

POST https://stenobird.com/v1/public/podcasts/arrested-devops/episodes/continuous-feedback-with-roni-dover/transcription-requests
Idempotently request low-priority transcript generation for this episode.
GET https://stenobird.com/podcast/arrested-devops/continuous-feedback-with-roni-dover.md
Read the agent-friendly Markdown representation of this episode resource.

Summary

Modern DevOps focuses heavily on deployment speed and cadence, yet often ignores the quality of the feedback loop. Roni Dover argues that without continuous, automated feedback during development, engineers are merely coding to pass tests rather than ensuring real-world performance.

Topics

DevOps
Continuous Feedback
Software Observability
Technical Debt
Software Development Life Cycle
Performance Engineering
System Thinking
Automated Instrumentation

Highlights

Main idea: DevOps processes are currently optimized for deployment frequency rather than the quality of the learning loop
Failure mode: Relying solely on post-deployment observability leads to engineers 'looking for trouble' in logs instead of preventing it during development
Practical takeaway: Implementing automated instrumentation and continuous feedback allows developers to see the impact of code changes before they reach production
Main idea: Observability data serves as an objective counterweight to the inherent biases developers have toward their own code
Failure mode: Without accessible performance metrics, developers default to 'coding to the test' rather than coding for actual user experience

Chapters

1:00 The Problem with Log Aggregators: The frustration of losing context while querying logs and the need for faster issue resolution.
4:45 The Optimization Trap: Why focusing on deployment cadence and release frequency fails to improve the software learning process.
8:35 The Data Gap in Development: Comparing the high level of data available to Product Managers versus the lack of actionable performance data for developers.
20:20 Accumulating Technical Debt: How the lack of continuous feedback leads to a cycle of troubleshooting instead of developing.
24:10 Countering Developer Bias: Using objective production data to overcome the natural bias of assuming code works as intended.
27:40 The Burden of Manual Analysis: Why engineers struggle to manually aggregate percentiles and performance metrics from raw data.
39:15 Practicing Observability During Development: Using automated instrumentation to provide developers with informed feedback during the coding phase.