Long-Running RPCs
When dealing with long-running RPCs, relying solely on a single connection can lead to significant data loss if the network fails. Instead, utilizing a pub/sub architecture allows clients to publish requests to a service, ensuring that results can be sent back without being tied to a single connection. This approach enhances reliability and prevents the loss of valuable processing time.In this clip
From this podcast

Software Engineering Radio - the podcast for professional software developers
Episode 421: Doug Fawley on gRPC
Related Questions