You can do this.

Let's say you have a 15 second trace interval. You run continuously, so you're collecting 5760 samples per day (4/minute, 60 minutes, 24 hours), or 40320 samples per week. Since you collect 240 samples in an hour, you can set up an alert to look at 40320 samples (the "Samples to examine" setting), and alert when 240 or more samples are lost (over 9999 ms).

That alert will look at the last 7 days, and alert you if the target has been down for more than 60 minutes total in that period - even if that 240 lost samples (or 60 minutes of downtime) is separated over 15 separate outages.

I've attached a screenshot of the alert configuration. This assumes a 15 second trace interval - if you're using something different, then you'll have to use different numbers - you should be able to just scale things proportionally, though.

Let us know if this is what you're after.

- Pete