Underneath the covers, SignalR wraps all communication between the server and clients into messages holding all the information with its origin, what the message is for, and the content of the message. By default, these messages are kept in memory in the process that hosts your SignalR-based solution.

This means that having two servers will not have inter-process communication going on, so one client sitting on one server and another on a second one would not know about each other's messages. With the flexibility of SignalR at the core level of it dealing with this through well-defined interfaces, it is fairly simple to make it scale out for different technologies. This is something the SignalR team has done as well; they provide the ability to scale out in different ways. You get support to use Microsoft SQL Server for temporary storage of messages between servers, or use Windows Azure Service Bus to distribute the messages, or even the popular Redis to...