This can be caused by routers that don't update their counters as often as PeakHour is set to query them. It results in a kind of "on" / "off" pattern in the graph, similar to this (sometimes worse):

There are two ways to fix it:

  1. Reduce the frequency the PeakHour polls. This is controlled in Display preferences via the Refresh Interval dropdown.
  2. Try enabling Ask Device to Smooth Data under Display options. This will send an snmpset command to the device each time PeakHour starts that asks the device to increase the frequency of it's updates.

    Support for this is device-dependent.

If none of the above options are suitable, we recommend you contact the device manufacturer and ask them to patch their SNMP implementation to allow for more frequent updates.