On Fri, 9 May 2003, Tace wrote: > >> CSMA/CD - Jamming Signal > I am not too sure about the presence of this jamming signal in > practical world, but I was taught it exist during my basic networking > course. The thing most closely resembling a "jamming signal" is found on twisted pair repeaters, where they inform stations of a collision with a jamming signal. This is different than shared media (co-ax) Ethernet where the receivers see a corrupted signal caused by a mix of the two transmitted signals (when listening) or a too-high voltage (when transmitting). With a repeater-based system the collision is represented by a (jamming) signal to all stations. A little tidbit that most people don't know: on a 10bT or 100bTx repeater a successful transmit results in your data on everyone elses Rx pair and _no_ signal your Rx pair. A collision results on a generated jam, not a mix of the 2+ signals, on everyone's Rx pair. These two rules, especially the former, mean that a repeater is more than just a signal amplifier: you can't fake it with simple components. It's difficult to know what you really want to measure, but you likely want to count collisions. All hardware will count transmitted OK failed due to too many (16) collisions Essentially all hardware will count Transmitted with no collisions Transmitted with one collision Transmitted with multiple collision (by exclusion) Most hardware will also count exactly how many collisions occurs. But this number is much less useful than it sounds. It is only useful with timestamps detailed enough to figure out what the deferral and collision-based delays really are, and those delay times are the directly useful values. But this is all moot: operational networks are almost exclusively switch based. With switches, the useful figure of merit is the percentage of time that flow control has been triggered. Alas, no NIC hardware provides that number. -- Donald Becker becker@scyld.com Scyld Computing Corporation http://www.scyld.com 914 Bay Ridge Road, Suite 220 Scyld Beowulf cluster system Annapolis MD 21403 410-990-9993 - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
