[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: automating the network: the big picture...





--On tirsdag, februar 11, 2003 12:53:17 -0500 Ralph Droms <rdroms@cisco.com> wrote:

I was also thinking about a slice through the problem...the minimal set
of information a host needs to have in order to communicate on a network:
definitional danger: "communicate on a network" is a term that people will disagree greatly about.

for example, if I am a device that does email client only, not knowing my SMTP server and POP (or IMAP) server will leave me unable to communicate no matter what else I know.

if the set you're looking for is the smallest possible set for all circumstances, it can be even smaller...

IP address
+ subnet mask
IP address of a router
not required if the medium supports ARP (bletch), is point-to-point or provides other means to "route" outgoing traffic appropriately

IP address for DNS service
insertion of the device name and address in  the DNS service
not required if the device is content to remain nameless

IP address through which other services can be found
(more?)

The idea here is to have a basis for the assessment of a protocol suite -
for example, in IPv6, we don't really have a good story for some of the
items in this list, yet.
I would group it, rather than have one single minimum:

- stuff you need to be able to send and receive IP packets at all
- stuff you need to be able to do IPSEC proof-of-who-you-are
- stuff you need to be able to get at configuration data
- stuff you need to be able to do function X
 (for instance, for mail you need your own email address(es), the 2
 servers, which probably means you want DNS because IP addresses are Just
 Too Ugly, a login identity for the servers, proof of that identity,
 authentication information for the servers (for MITM resistance).....?)

Another slice at the problem is the infrastructure that must be provided
by a network to support host configuration.  I'm trying to have a
framework in which to discuss the effects of new mechanisms that may
simplify the job of the host at the expense of causing more work for the
network infrastructure or of limiting the scale of the network
infrastructure.

- Ralph