Case Study: StartApp cuts compute costs by 33% with Granulate

How StartApp cut compute costs by 33% in 2 weeks

StartApp, a mobile data platform based in New York, needed to reduce rising compute costs as its business grew in the competitive ad-tech market. Its containerized Java web application, running on Tomcat and handling more than 100,000 requests per minute, was CPU- and network-bound, and the company wanted to improve margins without taking on expensive long-term R&D work. Granulate’s solution was chosen to help lower spend while maintaining performance and service quality.

Granulate deployed agents through Ansible, first in passive learning mode and then with active optimization for thread scheduling and memory allocation. After rollout across the cluster, StartApp saw a 30% reduction in average latency, 15% higher throughput, and 25% lower CPU utilization, which enabled the team to shrink cluster size and achieve a 33% reduction in compute costs in just two weeks, all without code changes.

StartApp

Motti Shpirer

Director of Operations