EXANTE has turned a routine engineering task into a lesson for the wider FinTech industry. When the broker split its order risk assessment service out of its core monolith, it found that AI could catalogue its tests in hours but could not be trusted to migrate them. The manual work that followed uncovered hidden flaws that had sat unnoticed for years.

The project was led from the quality assurance side by Grigorii Sedelnikov, an automation QA engineer at EXANTE. What first looked like a simple job quickly grew into an epic. EXANTE automation QA engineer Grigorii Sedelnikov said, “How hard can it be to split a monolith? You just move chunks of code into a new service, update the endpoints in the tests and you are done.” He admitted he once believed this too, before the team began finding duplicated tests, logic errors and uncovered cases.

EXANTE’s monolith, known as Core, is essentially a giant calculator. It processes orders, positions and margin settings for hundreds of thousands of clients across more than 3 million financial instruments, from shares and bonds to futures and options. Every other service in the platform sends requests to it.

The component EXANTE extracted, Pre-margin, assesses risk when an order is placed. It is triggered whenever a client simply hovers over the Buy or Sell button, which means it receives far more requests than orders actually executed. Inside the monolith, it could only scale as part of the whole system. Running it as an independent microservice lets EXANTE allocate resources to it directly and gives one team full ownership of its domain.

Source link