Given the difficulty of finding the optimal solution to this problem, I think I agree with the earlier suggestion to just monitor the situation and report to the user if there is a problem - and provide a useful account of how to handle it. I imagine it would not be too difficult to write cron scripts to monitor the count. Perhaps one could measure load cycles over the last hour, the last day and the last month, for example. There could be some sort of notification if some threshold or other was overstepped. At this point it is of course less straightforward - what sort of notification should that be? A 'normal' desktop notification would perhaps work for most users? Normally I wouldn't suggest something like this here for fear of getting my knuckles rapped - at present I really don't have time to do it myself - but maybe someone is just itching to get going on a little project like this.