Automation on GeekyRyan

Data Tier Architecture Modernization - Phase 1: Rethinking Legacy Orchestration

Mon, 15 Jun 2026 09:00:00 +0000

Phase 1 of modernizing a legacy application configuration orchestration system - the scripts and automation that provision databases, apply configuration, and run SQL changes across environments. The old system was a sprawl of individually-named scripts, each handling a slightly different variant of similar work. Hard to maintain, hard to onboard onto, impossible to add consistent error handling or monitoring. Often hard to understand.

Rewrote it around a small set of reusable, well-defined action types instead: configuration action (routing sensitive and non-sensitive config to the appropriate configuration store), database-creation action (with input validation), SQL-script-execution action (runs scripts with clear parameters), and a general-purpose command action for anything else.

Enabling Automated, On-Demand Lab Environments

Mon, 12 May 2025 09:00:00 +0000

Let people spin up lab environments on-demand, self-service, instead of having to ask an infra team and wait around.

It streamlines testing, experimentation, and onboarding - a new team member can provision a full application stack without waiting on manual provisioning, which removes friction on both ends: the people don’t have to wait, and the infra team doesn’t have to handle the recurring “can you build me an environment” requests.

Rebuilding a Lab Environment Deployment Pipeline From Scratch

Fri, 08 Sep 2023 09:00:00 +0000

Lab and test environments have a tendency to accumulate complexity over successive iterations of “just add one more special case.” By the millionth version of our lab deployment pipeline, it had grown into something that was hard to reason about and expensive to maintain - lots of administrative overhead just to keep it running, let alone extend it. Deploying the pipeline also took far too much time.

This project was a ground-up rebuild of that pipeline with the explicit goal of removing accumulated complexity and reducing the administrative burden of maintaining lab environments. Rather than patch the existing pipeline again, I rebuilt it with a cleaner, more streamlined deployment process - fewer special cases, more consistent conventions, and less manual intervention required to stand up or tear down a lab environment. The overall deployment time went from 45 minutes to less than 25 minutes.