Write-ups of systems that run in production: what was built, why it was built that way, and what the numbers showed.
9 min read
How SpendCentric keeps personal and organization data in the same tables, removes the inactive tenancy branch from every global query filter before the SQL is written, and lets SQL Server seek one index per mode.
14 min read
How SpendCentric moved account balances into a double-entry ledger, committed money effects with their source rows, and retired its event-driven balance pipeline.
13 min read
Moving payment processing out of web requests and onto GoodJob and Postgres, with the payment record as the durable intent, jobs that can be replayed by payment ID, and the numbers from a load test of the result.