How reliable is SignalR Backplane regarding to the question if all messages will reach all subscribed nodes? Is it using a reliable protocol underneath or are there chances that a message can get lost?
Obviously it can be that (for example) due to some network issues one node is down for some time. When it becomes reachable again, SignalR Backplane will deliver all intermediate messages. This is at least what I understand from davidfowl:
[...] This is VERY important! SignalR is NOT reliable messaging, it's a connection abstraction. We may buffer messages for longpolling but you cannot rely on the messages being there for ever. If you have important messages you need to persist, then persist them.
But how long is "forever" in this context? Can it be quantified/configured?
Are there other scenarios to consider if a reliable system is to be built on top of SignalR Backplane?