A Generative AI Engineer has provisioned a throughput model serving endpoint for a RAG application. They currently use a microservice to log incoming requests and outgoing responses to a remote server but are looking for a native Databricks feature to perform the same task. Which feature should they use?