Fee-calculation performance on a high-volume public-sector platform
Fee calculation ran for tens of seconds on the heaviest records and blocked the UI. Measured 3.4x to 15.4x faster.
- Result
- 3.4-15x faster
On a high-volume public-sector platform, fee calculation ran for tens of seconds on the heaviest records, which blocked the UI and the batch pipeline.
I went through every caller of the calculation engine. A recursive double-execution and per-row database writes accounted for almost all of the time. I removed the recursion, batched the writes with table-valued parameters, and rewrote the correlated subqueries in the stored procedures. I checked each change against real production data before rollout. The worst case I measured went from 25.6 seconds to 1.7.
More work
Migrating legacy systems into a platform that models the work differently
Close to two million records moved out of ten separate legacy systems, each from a different vendor with its own idea of what the data means. Getting the rows across is the easy part.
Business-intelligence dashboard for mid-market operations
A management dashboard that turns years of operational data into real-time financial-health, profitability and collections insight.