The prior eight papers in the HOWL infrastructure series specify what to build. INFRA-1 establishes the taxonomy. INFRA-2 specifies the OpsDB design. INFRA-3 demonstrates a schema. INFRA-4 specifies the runner pattern. INFRA-5 specifies the API gate. INFRA-6 specifies schema construction. INFRA-7 introduces the architecture to new readers. INFRA-8 specifies the shared library suite. None of them say how to actually start. This paper specifies the implementation path: a defined sequence of six phases that take a team from "we have read the specifications" to "we have a working OpsDB serving real operational data." Each phase makes a specific decision, produces a specific deliverable, defers what the next phase will address, and has a validation criterion that determines whether the team can move forward. The phases are: decide cardinality, determine baseline schema, build the development API and start ingesting data, determine the shared library core, design and implement change management, and add operational logic beyond OpsDB management. The structural claim is that the architecture is large enough that attempting to build it all at once produces a multi-quarter project that delivers nothing usable until the end. The phased path delivers operational value early and grows into the full architecture; each phase validates the team's understanding before committing to the next. What this paper does not specify: storage engine choice, programming language choices, deployment topology, specific identity provider integration, or implementation timelines. Those are organizational decisions that depend on the team's existing context.
Geoffrey Howland (Wed,) studied this question.