Your company has a Kubernetes application that pulls messages from Pub/Sub and stores them in Filestore. This simple application was deployed as a single pod. Upon analyzing Pub/Sub metrics, the infrastructure team found that the application cannot process the messages in real time, resulting in most messages waiting for minutes before being processed. The processing of these messages is I/O-intensive. How should you scale the application's elaboration process to handle the workload effectively?