When we ran services like this in-house, I don't really recall a time where any of them failed. Now that we have a 3rd party run those services, it's easy to recall multiple instances where one or more of them were down for some reason.