[comp.protocols.tcp-ip] X.25 service

dgolber@SM.UNISYS.COM (David Lawrence Golber) (02/17/89)

In article <14077@bellcore.bellcore.com> karn@jupiter.bellcore.com
(Phil R. Karn) writes:
>Nor did Telenet seem interested in fixing their
>switches to not act as though the D bit were always set.

The obvious thing to try: go to other vendors!  That IS one of the
advantages of a standard; there are lots of vendors - at least in
the US.  (When the only vendor is the government PTT, you may have a
problem.)

Nothing gets the attention of a vendor more than loosing business.

I don't think it's the 7-bit sequence numbers you want; it's the
act-as-though-no-D-bit is set.

I'd be real interested in what you find out about different vendors
... and I'm sure lots of other people would to.

karn@jupiter..bellcore.com (Phil R. Karn) (02/17/89)

>>Nor did Telenet seem interested in fixing their
>>switches to not act as though the D bit were always set.
>
>The obvious thing to try: go to other vendors!
>[...]
>Nothing gets the attention of a vendor more than loosing business.

We did! We junked the 9600 bps Telenet X.25 interface several years ago and
went with a 56kbps leased line direct to a port on the ARPANET. Costs far
less and works much better, although I did have to insist I wanted an 1822J
(HDH) port, NOT an X.25 port. It's still amusing to see all that HDLC link
layer overhead on the protocol monitor, but since the data frames are large
and the delays small, it doesn't create any real problem. And, of course,
our gateway is immune to "virtual cirkosis".

>I don't think it's the 7-bit sequence numbers you want; it's the
>act-as-though-no-D-bit is set.

Actually, either one would have been a help, since "local significance" in
the packet layer RRs would reduce the delays and therefore increase the
protocol-imposed throughput limit just as a larger packet layer window
would.

Please don't take any of my criticisms of Telenet and X.25 as criticisms of
the CSNET team at BBN. They did the best job possible given brain-damaged
network facilities, and the fact they could make it work at all can be best
described as heroic.

Phil