darrelj@sdcrdcf.UUCP (Darrel VanBuer) (01/04/85)
DEC has announced Ethernet tranceivers for use on broadband networks. They implement the standard 10 M bps in 18MHz of bandwidth. There are configurations for either single or double cable broadband. Maximum distance is 3800 meters. Available in the spring. The bad news is that a DECOM tranceiver costs $4250 (and the accessory for single cable broadband another $4500--but I can't tell from the press release whether that's per node or per net). By way of comparison, baseband Ethernet tranceivers cost as little as $200. The announcement also reminded the existence of the DELNI clustering device which allows up to 8 stations to share one tranceiver (a DELNI costs about $2000, so it clearly pays for a group of machines within 50 meters of a common point!). Comments: cost is negligible for supermini or mainframe systems, rather steep for machines in the PC class (it costs about $1000 to put an IBM PC on baseband Ethernet--and that includes the Ethernet controller and outboard tranceiver, $600 to $4000 more for broadband) -- Darrel J. Van Buer, PhD System Development Corp. 2500 Colorado Ave Santa Monica, CA 90406 (213)820-4111 x5449 ...{allegra,burdvax,cbosgd,hplabs,ihnp4,orstcs,sdcsvax,ucla-cs,akgua} !sdcrdcf!darrelj VANBUER@USC-ECL.ARPA
ian@mulga.OZ (Ian Richards) (01/07/85)
> DEC has announced Ethernet tranceivers for use on broadband networks. They > implement the standard 10 M bps in 18MHz of bandwidth. There are > configurations for either single or double cable broadband. Maximum > distance is 3800 meters. Available in the spring. > Darrel J. Van Buer, PhD Is the 3800 metres correct? If so, how do they do it? If they are still using CSMA/CD then the maximum distance between nodes is limited by the contention period (46.4 microseconds) which in turn dictates the minimum frame size (512 bits). The DEC/Xerox/Intel "standard" suggests that after you allow for all kinds of switching and hardware delays you have about 23 of the 46 microseconds left over for propogation time which at 0.7c is pretty close to 5000 metres. This of course is the "round trip" time which means a maximum node separation of 2500 metres. Now if you use a broadband cable surely this ought be halved again. This is because the worst case contention problem is two nodes near to each other but furthest from the head end. One begins transmission. Its signal propogates via the head end and all the way back until it almost reaches the other node which then begins a transmission. This signal (which is now a collision) must propogate all the way via the head end and back to the first station so it can detect the collision. The first station must continue transmitting for all of this time. Now perhaps they can avoid the double propogation delay by doing the collision detection on the transmission channel/cable. That brings it back to the baseband situation. (But how does a station upstream from a transmitting station find out that if it transmits it will cause a collision?) Perhaps they can get rid of some of those aforementioned delays or perhaps their signals go faster than 0.7c. I would be interested to hear the answer if anyone knows. Ian Richards decvax!mulga!ian
mccallum@nbires.UUCP (Doug McCallum) (01/14/85)
> Is the 3800 metres correct? If so, how do they do it? I am assuming that DEC followed the proposals put forth in the IEEE 802.3 committee for their implementation of a broadband CSMA/CD. The proposed addition of an AUI Compatible Broadband to the IEEE 802.3 standard allows up to 4km. for the length with 0.87c cable. There are other differences in the spec as well. > If they are still using CSMA/CD then the maximum distance between > nodes is limited by the contention period (46.4 microseconds) which in > turn dictates the minimum frame size (512 bits). The DEC/Xerox/Intel CSMA/CD doesn't impose the contention period, the "Ethernet" spec does. By changing many of the parameters to better fit the existing broadband media, it is possible to get a CSMA/CD system with a larger area, different data rates, etc. The IEEE 802.3 standard is making allowances for CSMA/CD on different media and data rates. > back to the baseband situation. (But how does a station upstream from > a transmitting station find out that if it transmits it will cause a > collision?) Perhaps they can get rid of some of those aforementioned The signal being used in the proposed broadband 10Mb system (proposed by DEC, M/A COM and possibly others) uses 3 CATV channels and splits these into a data channel and an outofband channel for collision detection. If two stations start broadcasting at the same time, collision can be detected by comparing what is received with what is transmitted. The comparison need only be done on the header through the source address. I really don't know much more than that. The above came from some notes I took at the last IEEE 802 meeting. Doug McCallum {allegra,ucbvax,ut-sally}!nbires!mccallum
hes@ecsvax.UUCP (Henry Schaffer) (01/14/85)
A DEC technical support person told me that the 3800 meters maximum distance means 1900 meters max from each transceiver to the Head End of the broadband. --henry schaffer