A system and method for dynamically allocating cloud resources includes receiving an application data request from a user and routing the request to a specific cluster of servers. Each application has a record in a database that contains a unique token along with subscriber information. Application requests to the service provider backend, are routed by DNS using an API endpoint name and originating geographic location to appropriate server clusters. When the request reaches the cluster, the load balancer extracts the application token from the URL, looks up the record in the database, and finds the subscriber plan and identity the application is associated with. A load balancer routes the data request to a specific cluster based upon the subscriber plan and identity. The clusters can have different processing speeds and features based upon the subscriber plan with premium plans having better performance than lower end plans.