leif@control.lth.se (Leif Andersson) (10/12/89)
I am posting this for a friend without net access. Please reply direct to him. /Leif ------------- I would like to know the experiences people have with implementing localtalk with different kinds of cabling. The original Apple cable is (of course...) prohibitively expensive. What are peoples experiences of junk cables? The middle pin needs not to be connected, right? Is some special cable characteristics important? Is it essential to have the shield connected? After all, should one by the cable from Apple? What can be done about the lousy quality of the connectors? Note that, obviously, for just a few nodes, "everything" works; the problem is when the network is growing. Our problem is like follows: We presently have a 12-node localtalk network, cable length not exceeding 100 meters, We have Mac II's, Mac SE's, two LaserWriters, one ImageWriter and one Kinetics FastPath 4.1. The latter connects the localtalk to an Ethernet, consisting of Unix boxes. It runs the KIP software, and on the Unix side we use the Columbia Appletalk Package 5.0. We use the file server Aufs, Telnet via the Kinetics, printing over Localtalk etc. We have severe performance problems with all nodes connected to the network. These problems look like follows: Everything shows up in the chooser, but the network performance is SOO slow that basically the net stops working. After disconnecting some nodes, it does not matter which, everything works nice and fast. Disconnecting different components reveals no single faulty node or cable. The official limits are 30 nodes and 300 meter cable, so we are far below. Can this depend on the junk cables we are using? Or anything else? Any help? I have tried different troubleshooting programs (Netcheck, look, etc) but not been able to spot the problems. Please respond to me by mail, since Usenet does not reach University of Bremen :-(. I'll summarize to the net. Bengt Martensson +49 421 218-2952 Institute for Dynamical Systems +49 421 171713 (home) University of Bremen FAX: +49 421 218-3054 Postfach 330 440 bengt@mathematik.uni-Bremen.de D-2800 Bremen 33 bengt%mathematik.uni-Bremen.de@uunet.UU.NET F.R.G. G09E@DHBRRZ41.BITNET