Jeff Kell writes:

>You're hitting pretty close to home here and I must respectfully
>disagree.  Appletalk version 1 has essentially an 8-bit address space,
>while version 2 allows a "cable range" of consecutive 16-bit network
>numbers.

Right -- a 16-bit cable range plus an 8-bit station ID is 24 bits.

>When we have a prolonged power outage, it can take a day or more to get
>Appletalk working again.  Typical Appletalk devices come up and listen
>for their network (cable-range) number.  If they don't hear one, they
>make one up.  Then they make up a node number and check for a conflict.
>The consequence of this is that the first AT device that powers up and
>sets its network parameters dictates the network number that will be
>used, and subsequent devices will follow suit.
>...
>Most of the time it works great, and I can appreciate the logic behind
>it for typical applications.  However, it is the most difficult to debug
>when things go wrong, speaking from experience.  We route AT across 12
>subnets.  I dread power outages :-(

Again, this is both the reason that an AppleTalk network can be
self-configuring, and the reason that the self-configuration won't work
past a certain size. When the self-configuration fails, it gets in the
way. Something has to set up network addresses, and since the AppleTalk
design assumes there's no administrator, devices use a statistical method
to assign addresses when they can't find an electronic authority. This
doesn't work when there's a human being trying to maintain order.

However...

>> >TCP/IP isn't self-configuring even at two nodes,
>
>One word (well acronym) ... DHCP

DHCP is not the same thing as self-configuration. First, someone needs to
be knowledgeable enough to (a) know that a DHCP server needs to be set
up, and (b) understand TCP/IP networking well enough to set one up.
AppleTalk requires nothing more sophisticated than being able to plug in
the cable. DHCP simplifies an administrator's life considerably, but  a
small AppleTalk network doesn't need any administration at all.

The reason I mentioned it at all is not because I think that AppleTalk is
a superior internetworking protocol -- it's not, and was never designed
to be. I mentioned it to point out that the complexity that Gavin
originally complained about should disappear in the future, as the
technology for managing networks catches up with growth far beyond
anything ever anticipated by the designers of today's network protocols.

-- Bruce


--------------------------------------------------------------------------
Bruce Toback    Tel: (602) 996-8601| My candle burns at both ends;
OPT, Inc.            (800) 858-4507| It will not last the night;
11801 N. Tatum Blvd. Ste. 142      | But ah, my foes, and oh, my friends -
Phoenix AZ 85028                   | It gives a lovely light.
btoback AT optc.com                |     -- Edna St. Vincent Millay
Mail sent to [log in to unmask] will be inspected for a
fee of US$250. Mailing to said address constitutes agreement to
pay, including collection costs.