mac@uvacs.UUCP (01/11/84)
I agree with (fortune.2005) and others that BiSync is a poorly specified disaster. The different levels of a protocol are all mashed into one big spec, which has never been written down. For example, transmission block boundaries are significant to the customer. Acks are numbered, but not messages, so the loss of certain control messages can hopelessly foul up the error recovery. The stations are asymmetric, the difference being in the waits on an idle line. The difference is on the order of a second, which AT&T threatened to introduce by using satellite links for leased lines. Face it, we've learned a little about communications protocols in the past ~15 years. The only virtue of BiSync is its near-universal availability. We used it to make a DTSS (Honeywell) mainframe appear as a RJE station to COMTEN front ends on a Amdahl 470 running some IBM system, as well as to allow Wang mincomputers to appear as terminals to DTSS. The drawback is that all parties involved had different ideas of the protocol. Furthermore, there are many flavors: 2780; 3780; 327n, with or without transparent text, block transparency, ad nauseum. We also ran into trouble with differences between X.25 networks ({Tele,Tym,Uni}Net,{Trans,Data}Pac, etc.) but nowhere near as messy. Alex Colvin ARPA: mac.uvacs@csnet-relay CS: mac@virginia USE: ...uvacs!mac