A leading global mobile data platform, Start.io reaches over 1.5 billion monthly active mobile users worldwide by partnering with more than 800,000 mobile applications. Start.io leverages the power of mobile data to explore user behavior and intent to reach targeted audiences on a global scale.
Following growth in the Asian market region, Start.io implemented a series of internal infrastructure changes to tackle rising compute costs. Start.io was looking to further reduce their compute spend without getting involved in expensive long-term R&D projects.
After a 24-hour learning process, Intel Granulate was activated with capabilities including thread scheduling and memory allocation. The optimization solution adapted to the latency breakdown, automatically reprioritizing threads and taking into account each request’s processing stage.
Following Intel Granulate’s activation on the entire cluster, Start.io achieved 30% reduction in average latency, 15% throughput increase and a 25% CPU utilization reduction. Since Start.io’s priority was to maximize cost reduction, they began by reducing cluster size and removing machines until reaching pre-Granulate SLAs and achieved an astounding 33% compute cost reduction.
30% Reduced Average Latency
Figure 1: Reduced latency with Intel Granulate (blue) vs. without (red)
Improved RPS
Figure 2: Improved RPS with Intel Granulate (blue) vs. without (red)
25% Reduced CPU Utilization
Figure 3: Reduced CPU utilization with Intel Granulate (blue) vs. without (red)
*Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy
Start.io turned to Intel Granulate's real-time continuous optimization solution to slash costs. Following a quick deployment of Intel Granulate's solution, Start.io achieved dramatically improved performance while reducing compute costs by 33%.
