A paper I skimmed.
It is about error in measurements done in a switch. There are four properties discussed:
- Timing. The timestamp of packets traced.
- Loss. The packets missed by the packet capture.
- Gain. The packets did not exist but captured. These packets are called phantoms by the paper. Gain occurs because switch may replicate a packet when forwarding it. For example, when a packet flood to more than one port, and the measurement is done on multiple output ports, more than one event are recorded for one packet forwarding event.
- Layout. Identification of topological components, such as whether traffic is intra- or inter-subnet; or the broadcast domain of an IP subnet; or detecting hidden switches.
Bibliographic data
@inproceedings{
title = "On Calibrating Enterprise Switch Measurements",
author = "Boris Nechaev and Mark Allman and Vern Paxson and Andrei Gurtov",
booktitle = "Proc. IMC'09",
month = "November 4-6",
year = "2009",
address = "Chicago, Illinois, USA",
}