North American Network Operators Group
Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Re: redundancy [was: something about arrogance]

  • From: Pedro R Marques
  • Date: Wed Jul 31 02:39:34 2002 
Pedro Roque Marques wrote:
------- Start of forwarded message -------
From: [email protected] (Patrick Evans)
To: Jim Shankland <[email protected]>
Cc: [email protected]
Newsgroups: jnx.ext.nanog
Subject: Re: redundancy [was: something about arrogance]
Message-ID: <[email protected]>
Date: 31 Jul 02 00:32:49 GMT
References: <[email protected]>
Organization: Juniper Networks, San Francisco, California


On Tue, 30 Jul 2002, Jim Shankland wrote:


Patrick Evans <[email protected]> writes:


My first project, if network availability were a key issue, within any
organisation would be to a) obtain [an AS number] and b) make use of
it.

Heh. How many bits in an AS number, again?


*grin*

That's a problem with the underlying protocol. I get paid to run
operational networks, not bleat endlessly about "how much work would
it *really* take to implement 24bit AS numbers?" :)

The plan is 32 bits... (see draft-ietf-idr-as4bytes-05.txt for details).
Essentially i think it just takes interest/demand from ISPs since the mechanism can be implemented and deployed without in a non disrruptive way.

Crying about protocol deficiencies is a distant second to keeping a
business up and running these days.

imho, protocol efficiencies are not so much the problem. If it is clear the scale routing must operate on the right hardware/software can be engineered... that assuming that people are willing to upgrade their existing boxes and that it isn't a requirement that it must run on 5 year old small entreprise boxes.

The later seems to be the biggest problem although. Effectivly the growth of routing table size is bound by the maximum memory size and CPU capacity present in the most common boxes used in the network and not by protocol efficiency.

It is not so much of a question if one can build a database engine and respective distribution protocol than can scale upto n million paths but of the limits of the current day moral equivalent of the AGS+. Thus all the people that have these deployed in their networks tend to be concerned about the need to upgrade them as the size of the routing table increase.

As one of the posters was king enought to point out these sometimes end up being more issues of economics/buisiness than of engineering.

regards,
Pedro.