Production is the live environment where an engineered system serves real users or business processes. Changes deployed there can be judged by business impact rather than offline performance alone. section 1.0 section 1.1.3
Production is the final stage of an experimental improvement loop:
- Implement a code change and apply software quality controls such as code review and unit testing. section 1.0
- Evaluate the change offline using historical data or test datasets to reject clearly harmful or slower alternatives. section 1.0 section 1.1.3
- Deploy the surviving change to production and measure a business metric against the previous system. section 1.0 section 1.1.3
- Keep changes that improve the metric and reject changes that do not, repeating the loop for progressive improvement. section 1.0 section 1.1.3
Search server transformation
A software engineer suspects that data transformation slows responses. They time a replacement offline on previous request datasets, deploy it to production, and compare user bounce rate with the old implementation. A lower bounce rate justifies retaining the change. section 1.1.3
Figure 1.1 Common engineering workflow. (1) A new idea is first implemented as a code change to the system. (2) Typically, some offline evaluation is performed that rejects ideas that are expected to negatively impact business metrics. (3) The change is pushed into the production system, and business metrics are measured there, online. Accepted changes become permanent parts of the system. The whole workflow repeats, creating reliable, continuous improvement of the system.
Trace the complete progression from code change through offline screening to online measurement and permanent acceptance. section 1.0
