Commonwealth of Virginia Agency Systems
Written alone. In production for more than ten years.
- 0State agencies and offices
- 0Years in production
- 0Developer
The problem
Virginia's state agencies needed account management and statewide outage tracking that actually reflected how incidents were handled — not how a specification imagined they were handled.
Why it mattered
When an outage crosses a hundred agencies, the coordination is the product. Bad tooling there does not produce an inconvenience; it produces a hundred organizations operating on different information during an incident.
The old process
- Account management handled through fragmented manual procedure
- Outage tracking and communication assembled ad hoc, under pressure
- Procedural knowledge held by whoever happened to be on shift
- No standardized statewide reference for how any of it was done
The idea
The person answering the phone at three in the morning already knows what the software has to do. That person should write it.
The system
Full-lifecycle web applications for state agency account management and statewide outage tracking. Concept, data model, application layer, deployment, SME documentation and support — every line mine.
How it works
- Data model designed around how incidents and accounts actually behave
- ASP.NET application layer over SQL Server, deployed on IIS
- Used across 100+ state agencies and offices in daily operation
- SME documentation authored alongside, standardizing the procedure statewide
- Supported in production by the person who wrote it
Design decisions
- Requirements from ownership
- I was the Critical Outage Coordinator. The requirements were not gathered from a stakeholder; they were the job I was already doing at three in the morning.
- Documentation as deliverable
- The SME documentation standardized the procedure statewide. Software that changes what people do needs the written procedure to change with it.
- Built to be supported
- I knew I would be the one supporting it, which is a remarkably effective constraint on architectural cleverness.
Controls
- Standardized statewide procedure documented and published
- Structured incident communication across 100+ agencies
- Multi-shift operational coverage for 24/7 uptime
Testing
Production use across a hundred agencies for a decade. It is the least controlled and most conclusive test there is.
Result
Those systems ran for more than ten years. Software written by one person, holding up under a hundred agencies' daily use for a decade.
Impact
- Account management and outage tracking across 100+ Virginia state agencies
- Statewide incident communications, including NCIS, VAI and state security offices
- SME documentation that standardized procedure statewide
- Multi-shift team management for 24/7 uptime
- More than ten years of production life
What I took from it
Longevity is a design outcome, not luck. Systems last when the person who built them also had to answer for them, because that person makes different decisions about error handling and documentation than someone shipping to a handoff.
What I would build next
Nothing. It ran for a decade and that is the correct ending for a piece of infrastructure.