mcneill@eplrx7.UUCP (mcneill) (01/26/89)
One machine on our LAN caused a lot of havoc. It was mistakenly placed on
the network before it's networks file had been configured correctly. It
thought that it's network address was 192.9.200.0. I seem to remember
seeing somewhere "never use x.x.x.0 address" (where x ranges from 0 to
255). Is my memory correct? If that is an untouchable address why is it?
thanks
--
Keith D. McNeill | E.I. du Pont de Nemours & Co.
eplrx7!mcneill@uunet.uu.net | Engineering Physics Laboratory
(302) 695-7395 | P.O. Box 80357
| Wilmington, Delaware 19880-0357acmb@RELAY.PROTEON.COM (01/27/89)
One machine on our LAN caused a lot of havoc. It was mistakenly
placed on the network before it's networks file had been configured
correctly. It thought that it's network address was 192.9.200.0. I
seem to remember seeing somewhere "never use x.x.x.0 address" (where x
ranges from 0 to 255). Is my memory correct? If that is an
untouchable address why is it?
Keith D. McNeill | E.I. du Pont de Nemours & Co.
Keith,
My part of the history of that comes from MIT LCS and the initial work of
ProNET-10 that was done as v2lni. The format of the token ring had a
sensitivity to bit dropping (in the Manchester code) that caused a token to
become a beginning-of-message which would cause all machines in the ring to
experience an interrupt. That was clearly undesirable so we concluded it
should not be used as an address. The bit dropping was related to what the
cables do to differential Manchester encoding. There were other reasons
discussed at the time, but I think one of them was that some interfaces had
intended to use 0 as the broadcast address and was not to be used for
anything else. We had initially used 0 as broadcast and with this problem
cause discovered we concluded that 0 was VERY bad for broadcast and
inverted the bits to give 255 (0xFF) for the broadcast.
Those discussions took place at MIT about 1980-81 and the removal of one
more address from those used out of 255 was not expected to be missed by
anyone. Thus the recommendation is to not use address 0 for any node and
this carries over to other networks that use 8 bit addressing.
-Alan Marshall, VP Proteon Inc.
2 Technology Drive CCMAIL: acm at proteonwebo
Westboro, MA 01581-5008 ARPANET: acm@Proteon.com
tel: (508)898-2120 MHS: acm @ ProteonW