roy@phri.UUCP (Roy Smith) (09/08/86)
We have more devices in one room than we can plug into the ethernet
and still keep proper spacing on the tranciever taps. A DELNI (or similar)
device seems like the answer. Do people who have DELNIs have any complaint
with them? Are there other manufacturers that make similar products that I
should be looking at? If anybody knows of a DELNI-type box that handles 4
devices instead of the usual 8, I'd like to hear about that too (assuming
it costs less than an 8-device box).
--
Roy Smith, {allegra,philabs}!phri!roy
System Administrator, Public Health Research Institute
455 First Avenue, New York, NY 10016phil@amdcad.UUCP (Phil Ngai) (09/11/86)
In article <2429@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: > > We have more devices in one room than we can plug into the ethernet >and still keep proper spacing on the tranciever taps. A DELNI (or similar) >device seems like the answer. Do people who have DELNIs have any complaint >with them? Are there other manufacturers that make similar products that I >should be looking at? If anybody knows of a DELNI-type box that handles 4 >devices instead of the usual 8, I'd like to hear about that too (assuming >it costs less than an 8-device box). We have some DELNIs and I'd like to make the following points. 1) They do heartbeat wrong. You would assume that a DELNI should imitate a set of real transceivers as much as possible. As such, you would be able to select whether or not a port did heartbeat. You would never expect to get heartbeat when you hadn't transmitted anything. Of course, other devices on the coax which use receive mode collision detect could not receive a collision between two ports on the DELNI. Receive mode collision detect is based on the fact that two MAUs are transmitting and thus the signal amplitude is twice the nominal value. A DELNI has no way to signal this information. All it can do is transmit a jam. I can not think of any situations where this causes a problem. The DELNI is not configurable for heartbeat. If the associated transceiver does heartbeat, everyone gets it. If the transceiver doesn't no one gets it. This will become even more of a burning issue with the recent change in the 802.3 standard deleting heartbeat for repeaters but requiring it for all other controllers. Without port by port configurability, you will not be able to mix repeaters with other controllers on a DELNI. Also, sending heartbeat to someone who wasn't even transmitting is wrong. But that's what the DELNI does when connected to a transmitter which does heartbeat. Heartbeat is sent to everybody. As a hardware engineer who has talked to other hardware engineers, we believe it would be easy to remember who did the transmitting and route the heartbeat to that port. The DEREP is Ethernet 2.0 and therefore wants to see heartbeat at the end of its own transmissions. If, however, you send heartbeat to a DEREP which wasn't transmitting, the DEREP will jam the other side after every packet transmitted on the DELNI it is on. But that doesn't matter because: 2) DELNIs don't support DEREPs. We connect our buildings with fiber optic remote repeaters. All the fibers run into one hub location and we have a rack full of repeaters. A natural application for a DELNI. But DEC doesn't support this so we have to mount a transceiver farm on the wall behind the rack full of DEREPs. DEC doesn't even know why it doesn't work. But they tell you not to do it. (If you ask someone who's technical enough. Be sure the person supplying your information knows what they're talking about.) 3) DELNIs can't be rack mounted. In our environment everything is rack mounted. Unfortunately DEC thinks we should buy lots of tables to put their equipment on. There was a DELNI rack mount kit but it got killed. I think this is a mistake. I think people who agree with me should complain to DEC. The product marketing manager's name is Graham Slieker. You can send email to him at decwrl!slieker@derep.dec.com or slieker@derep.dec.com Yes, his machine really is named derep. 4) DELNIs can't be cascaded in CONNECTED mode. The _Networks and Communications Buyers Guide_ lists the three supported modes: standalone, hierarchical standalone, and connected. Hierarchical connected mode is not supported. It may work in some cases but it is not supported. Running red lights works in some cases but is not recommended either. This is not a complaint, just a point of information which many people have missed. Another little known fact is that transceiver cables lengths add when going through a connected DELNI. If the controller is good for 40 meters and it's 15 meters from the DELNI to its transceiver, you only get 25 meters from your controller to the DELNI. TCL claims their multiport can be cascaded. I do not know if they have a better design or if they are simply more willing to take risks than DEC is. BICC has a multiport. It sounds like a generic multiport. Be aware that BICC doesn't seem to follow the policy of "be liberal in what you accept and strict in what you send", for their 802.3 repeater doesn't work well with MAUs that send heartbeat. I don't know if there are any gotchas with their multiport. As for your four port request, you can do that with TCL at a cost of $841.00. By comparison, an eight port costs $1240.20. It is a big awkward box compared to the sleek DELNI. And you can't get much technical information on it. There is no technical manual for it. No specs are published for prop delay, squelch time, signal jitter or other such interesting parameters. It does have the advantage of allowing you to intermix Ethernet versions. -- Rain follows the plow. Phil Ngai +1 408 749 5720 UUCP: {ucbvax,decwrl,ihnp4,allegra}!amdcad!phil ARPA: amdcad!phil@decwrl.dec.com
msd@minnie.UUCP (Marc S. Dye) (09/12/86)
We have experience w/ DEC DELNI's here at NRC. They seem reliable.
And as you mention, they are great for turning many snakes into a
single octopus (no Sushi jokes, por favor). However, the ones we
have are *NOT COMPATIBLE* with some controllers which use a 'heartbeat'
signal (e.g. Excelan 20x). An alternative
apparatus which is similar which *DOES* work w/ Excelan controllers
is manufactured by TCL. If you want more information on that device,
send mail and I will dig up contacts, etc.
Marc S. Dye: USENET -> ihnp4!nrcvax!msd
{sdcsvax,hplabs}!sdcrdcf!psivax!nrcvax!msd
ARPANET -> ihnp4!nrcvax!msd@UCBVAX.BERKELEY.EDU
+----------------------------------------------------+
| *BADGES*? WE DON'T NEED NO STINKIN' BADGES!!! |
+----------------------------------------------------+