Umeå universitets logga

Financial trading systems require message pipelines that offer low latency, high throughput, and strong delivery guarantees. This thesis investigates how Apache Kafka, a distributed event streaming platform, can be integrated with the Financial Information eXchange (FIX) protocol in a cloud-native environment to meet these requirements. A prototype inspired by Nasdaq's Drop Copy service was implemented, with Kubernetes used for orchestration and deployment. Two Kafka pipeline variants were developed: one using a single partition to enforce global ordering, and another using multiple partitions keyed by financial instrument to preserve per-symbol sequencing. Both setups were evaluated under increasing message rates and simulated broker failures. The results show that Kafka can be configured to uphold the ordering and delivery guarantees expected by FIX. In failure scenarios, both pipelines maintained correct message handling, with only minor loss observed when all brokers were temporarily unavailable. These findings indicate that Kafka can serve as a reliable foundation for regulated financial messaging systems, while also identifying opportunities for improvement in failure recovery and scalability.