How reliable is the SignalR backplane regarding the question of whether all messages will be available to all signed nodes? Does it use a reliable protocol or is it likely that the message might get lost?
Obviously, this can be (for example) due to some network problems when one of the node does not work for a while. When it becomes available again, the SignalR signal board will deliver all intermediate messages. This is at least what I understand from davidfowl :
[...] It is very important! SignalR is not a reliable messaging, it is an abstraction of the connection. We can buffer messages for longpolling, but you cannot rely on messages that are there forever. If you have important messages that you need to save, then they are saved.
But how long is “forever” in this context? Can it be quantified / tuned?
Are there other scenarios to consider building a reliable system on top of the SignalR backplane?
Dejan source
share