Umm.. what *exactly* is upstream of that interface? I strongly suspect that it's *heavily* influenced by *local* preferences/configuration.
:^)
Care to speculate on the existence of any measurement point in the internet that is not heavily influenced by characteristics logically or physically local to it? Making statements about traffic in the network is a little like asking the temperature of the ocean; the answer depends a lot on who is asking and where the answering person is standing. That's why I fairly carefully referred to measurement points that *I* can read.
But frankly, every sample I have seen in the past decade placed SMTP in the "noise" category, a single digit percentage or less. Estimates I have heard from ISPs and seen on measurement sites available to me have consistently placed HTTP/HTTPS at better than 2/3 of total traffic in the late 1990's, and in the past few years have either estimated HTTP/HTTPS as neck-and-neck with file sharing traffic or have put file sharing well ahead. I recently understood from a Scandinavian ISP that DirectConnect is ~85% of their average workload.
Perhaps you could reply privately with your data; it would interest me.
_______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf