So, er... is there anything we can do about this?
And that includes the possibility of paying for better service from DeRoyal.
First off, the wording of the 508 error is misleading, and by the way error 508 does not even exist as a standard HTML code. The problem is not bandwidth, it's simply 100% CPU usage on our server. This happens routinely (every Sunday) for a few seconds when Witt and Wiki are backed up.
Regarding yesterday's lengthy outage, it was caused by a number of identical processes clogging up the resources. These processes are all instances of the php-cgi parser, so I assume that something, somewhere, is freezing the parser. (Note that you can't see this through cPanel, you have to manually SSH in.)
There is an "easy" temporary fix, until we troubleshoot further - simply schedule a killall -9 "php" as a cron job, every few minutes. Since normal php requests take less than seconds, you would be very unlucky to kill a legit request (and I don't think it would be a huge hassle if it happened), while you should easily clear away any stuck processes.