I believe the intention is that "peak" is not maximum, it's a snapshot of traffic during regular business hours, rather than all 24 hours. You can adjust reports to filter out non-peak (non-business hours) traffic:
Ok, lets try and break this down.
So, lets say you have an interface on a node, and that interface is polled every 10 minutes. If there was 600Mbits of traffic going across that interface during that period, it would take that 600 Mb, divide it by the time period (which it probably needs in seconds rather than minutes) and give you that result. So 600Mbits total traffic / (10 minutes * 60 seconds/min) = 1Mb/sec of traffic. If that is the highest value so far to day, it would be considered the peak.
Polling intervals can GREATLY affect this! Let's say that all of the 600Mbits of traffic shown above were in the 1st minute, with a 10 minute polling interval, the end result of 1Mb/sec would not be affected. However, if the polling interval was reduced to 1 minute, the equation would be greatly changed. ie: 600Mbits / (1 minutes * 60 seconds/min) = 10Mb/sec of traffic. Or more accurately 10Mb/sec for the first interval and 0Mb/sec for the next 9 intervals.
The maximum value in the given time period (ie: today) for any given period, would be the peak value for that day.