Scaling gRPC Applications via DNS Load Balancing

Having a single server that clients connect to over the network is acceptable if there are relatively few clients trying to connect to it simultaneously. However, what if our application needs to support thousands, or even millions, of connected clients? In this case, having a single server would no longer be sufficient. We would need to scale our application across multiple servers.

Scaling out is a process of deploying many duplicate server-side applications. These application instances still act as a single distributed application, but client connections are distributed among many instances, so each individual instance doesn't have to deal with more connections than it can support. The process of distributing the load between the applications is known as load balancing.