I have been witnessing some weird graphs on my graphs for the MRTG of my cisco router 2600 gateway. I feel its caused by network outages and lost pings from the MRTG server to the cisco router. I still feel it is so and when that happens i feel it get funny readings and measurements and tend to give a traffic value of nothing more than the speed of the interface it is querying e.g on an interface of 10Mbit with a bandwidth of 128Kbit to monitor, when the outage occurs, it can be giving values of 4,5,6,7,8 Mbit on both inbound and outbound. This distorts the graph analysis and makes planning difficult.
I dont know if u can give more accurate explanation of this and any remedy to it. For references, u can check the graph @ http://188.8.131.52/cisco.html