As a Google Professional Data Engineer, you are tasked with providing new website users a globally unique identifier (GUID). The GUIDs are generated by a service that processes various data points sourced from both internal and external systems. The data retrieval from these systems occurs through HTTP calls made by microservices within your pipeline. Given the high volume, with tens of thousands of messages per second that can be handled in a multi-threaded manner, you are concerned about system backpressure. What design approach should you take for your pipeline to effectively minimize backpressure?