The Human Scalability of "DevOps"

DevOps works very efficiently in small engineering organizations, but without careful management it causes serious human/organizational scaling problems in large organizations.

The author defines DevOps not merely as the fusion of development and operations, but as a model in which developers are made responsible for 24/7 operation of the production environment. This includes infrastructure management, reliability engineering, monitoring, incident response, and performance analysis, among other things.

The operation of internet applications has evolved through the following three stages.

• Stage 1 (Shrink-wrapped): The period when development, QA, and operations were separated, resulting in long engineering cycles.
• Stage 2 (Growth era): The emergence of automation and centralized infrastructure teams (SRE, etc.) led by companies like Amazon and Google.
• Stage 3 (Cloud-native): A structure that leverages public clouds so that developers handle operations as well, without dedicated operations staff.

Companies in the cloud-native era tend not to maintain dedicated operations staff, for the sake of labor-cost efficiency. However, this risks a lack of irreplaceable, skilled skillsets such as system debugging, performance profiling, and operational intuition.