Smarter Cloud Routing Cuts Costs 24% Without Slowing Apps

When you open an app, your request travels through an API gateway—a kind of digital traffic cop that connects you to the right server. For global apps, companies must decide which data centers to use. Put gateways everywhere: fast but expensive. Put them in one place: cheap but slow. This paper tackles that tradeoff.

Researchers built a mathematical model that decides how many gateway replicas to run, where to place them, and how to divide incoming traffic. Their goal: keep response times under a set limit while spending the least money. They tested it on simulated scenarios with 10 candidate regions. The optimal solution saved 24.2% monthly cost compared to running gateways everywhere, and it still met speed targets.

Because the exact formula is complex, they also created a shortcut algorithm. It ran hundreds to thousands of times faster than the full model, with costs just 3 to 5% higher than the ideal. That speed makes it practical for companies that need to adjust quickly as traffic changes. Picking the single cheapest data center was slightly cheaper, but it failed to meet speed requirements for 3 out of 10 regions—so balancing cost and latency matters.

In plain terms: this gives cloud engineers a smarter way to spend less money without annoying users with sluggish apps. For everyday people, it means better-performing apps and potentially lower subscription fees, as companies' cloud bills shrink.