Quotix, a fintech company specialising in intelligent trade execution and management for the FX industry, has implemented Azul Systems' Zing in order to eradicate Java garbage collection (GC) issues and latency outliers.
Daniil Volkov, Quality Assurance Lead at Quotix explains: "Quotix regularly runs various tests to study system behaviour during high volatility and identify system limits. During the stress-tests, which simulated extreme market conditions, we occasionally saw inconsistent trading execution. There were two reasons for this; the first one is a "warm-up" problem when the JVM would stop to optimize some part of the code at runtime, therefore delaying the execution. The second problem is that the JVM would pause to clean up memory using its garbage collection process. Both problems are common for Java-based software. Although the simulation was significantly tougher than we have ever seen in production, both in terms of market data updates and orders, it was considered a possible risk. Azul's Zing was selected as a technology that could help us eliminate it."
Azul Zing is a unique JVM that provides consistent low-latency with high throughput, supports large in-memory datasets without performance penalties, and solves the Java "warm-up" problem. Zing eliminates the operational interruptions and response time outliers often experienced by Java-based trading, risk and compliance applications in the financial services market.
Quotix's tests proved that Zing substantially reduces maximum pauses and enabled its FX aggregators to perform at higher loads. "The benefits of Zing over the JVM we used before were obvious. We have moved all critical Java components to Zing." Mr. Volkov continued.
Evgeny Sorokin, CEO of Quotix, said: "We implemented Azul's Zing to armour the Quotix backend. Our tests demonstrated that our systems powered by Zing can comfortably handle sharp spikes in the load and deliver consistent trade execution. It was very interesting to see how our solutions behaved during the recent CHF Black Swan. We saw several thousands of orders per second on each aggregator that were processed with no delays whatsoever. This is a great example of why we chose Zing."
Scott Sellers, CEO and President of Azul Systems, said: "Quotix handles tens of billions of dollars of transactions per day. Their business requires a Java runtime which can guarantee consistent real-time trade execution, even during peak trading periods and unexpected surges in volume. Only Zing delivers the high sustained throughput and consistent low-latency required to be competitive in today's real-time trading markets."