[I changed the Subject line...]
Date: Sat, 16 Feb 2002 17:50:18 +0200 (EET)
From: Pekka Savola <[EMAIL PROTECTED]>
Message-ID: <[EMAIL PROTECTED]>
| I also think DHCP and what not should not be used for router
| configuration.
I'm trying to understand what is lurking in the background here, and
I think I'm failing. That is, I have no idea at all what the objections
are, the oft repeated "I don't think ... should be used" (I picked just
one example above) means nothing - there's no explanation for what the
problem might be.
It could just be that people believe that the only way to ever configure
a router is manually, and that any protocol to automate the process is
necessarily flawed. If so, I'd like to understand why, but I think I'd
tend to ignore that sentiment - other than to make sure that implementations
keep on making it possible to manually configure, so that people who
believe this don't have the rug yanked out from under them (and in certain
particular circumstances, I think this probably includes all of us).
Or, it could be that there's something in particular about DHCP that makes
it unsuitable - its lack of authentication for example (is that still true
of DHCPv6? One day I must go read that draft...) - or that if you're going
to have to configure authentication manually, you may as well just config
the addresses (etc) instead.
So, could all those objecting please say exactly what it is that is the
problem, and not just repeat the "I don't like it" comments?
Then perhaps it will be possible for others to refute the alleged problems,
or to learn about a defect they weren't aware of, and fix the protocol (or
even invent a new protocol, if the existing possibilities are beyond fixing).
I find it a bit difficult to believe though that there's anything so
special about a router that makes it impossible to configure using a
protocol ... after all, we have (most of us) been using TELNET to configure
routers for many many years now, and while there's usually a human on the
client end of the telnet connection, there doesn't need to be - a script can
configure routers just as easily (and I've seen scripts around for this kind
of purpose).
These days routers even offer http servers as a config device, and automation
to talk to http servers is as comon as dirt. And then, there are also routers
that simply invent their own configuration if they can figure out some data
somehow.
So, it is pretty clear there's a demand for configuring routers some way other
than manually, it is also pretty clear that it is possible. If it can be done
and people want it, it sounds to me as if a standardised method (as an
optional extra, not as the only way) is likely to be an improvement.
kre
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
