Deployment Cost Optimization For Real-Time Risk SaaS Services In The Cloud

Sammanfattning

Microservice-based applications thrive when each service performs a specific, well-defined task independently, contributing to the overall system. However, what happens when these independent services overlap? Could consolidating some of these services yield substantial benefits? This thesis delves into the securities overlap among services within the Nasdaq Risk Platform. By evaluating metrics such as memory consumption, JVM overhead, and CPU utilization, it identifies potential candidates for consolidation. A proof-of-concept consolidation of actual Nasdaq Risk Platform services was deployed and compared against a non-consolidated version. The analysis revealed that the consolidated version performs similarly but exhibits a more uniform and stable usage curve. Additionally, by reducing the number of JVMs and combining the service securities cache, the total RAM usage decreased by 28%.