Re: Evolution and the routing architecture

From: Tony Li (tli@procket.com)
Date: Sun Apr 07 2002 - 01:38:48 EST


 | The real point I was trying to make is that one of the
 | requirements for the new architecture is that it should be
 | designed to be adept at adapting to changing conditions and
 | requirements.

On this, I would have to agree. An architecture should be flexible.

Our predecessors did a fine job of this and we have permuted (perverted?)
their work many times over to good results. Examples have been subnetting
and then CIDR.

One of the key properties that made this possible was the ability to
compartmentalize different aspects of the design and replace/upgrade/modify
them over time. Thus, the IP address that originally indicated your IMP
port became an independent network number and then became a subnet and then
became a prefix.

To generalize this, the point is that the components of the architecture
should be cleanly partitioned into subsystems. These subsystems
must be as loosely coupled as possible and that it must be possible to
replace subsystems incrementally.

An obvious question then is "what are the appropriate subsystems?"

Yes, I realize that this is motherhood and apple pie (glog? ;-), but I
think we should all be on the same page...

Tony



This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:04 EDT