Well if you don't want a single point of failure there are other methods that you can use such as several redundant trackers. The code hits the primary one and if it finds it ok it updates it and then the tracker informs however many backups that it has been told about. If it doesn't find the primary one it fails over to another down the list and so on. Then when the primary ones come back online they get the current state from the backups. While you may not care which sub-server handles the request, you may care if your broadcast gets dropped heaing out from the machine and the new subserver is placed under the impression that it is alone.
