At 10:59 AM -0400 5/1/02, Chris Charlebois wrote: >First of all, the idea of my washing machine having a globally routable >addess is a little scary. Someone could hack in and ruin my delicates.
You've just touched on a very real problem. On the one hand, it might be nice, on a hot day, to delay turning on your washing machine until it's cool, or do things like turn on your air conditioning an hour before you get home. There's been an interesting application in New York where daycare centers give parents webcams so they can check up on their kids during the day. On the other hand, there are significant potential security and privacy issues here. By and large, with the exception of IPsec, these are more application problems than routing problems. > >Second, in terms of waste, I understand what you are talking about when you >bring up the old "640K" arguement. I remember reading an article 10 years >ago saying that the 486 processor would never see the desktop, because it >was too powerful for anything other than servers. However, 128 bits *is* >alot, enough that you could take all the publicly routed IPv4 addresses, and >assign all of them to each square meter of the Earth's surface. Each square >meter (and that includes water) could be assigned a full 2^32 address >space. Until we start talking about extraterrestrial internets, I think >that 128 bit will do. You're quite correct in that it would be possible to assign addresses in that density. Of course, without massive geographic summarization, we could never route that many addresses -- BGP, as we know it, is reaching its limits (maybe 4-7 years at the present growth rate). Again, I'll reiterate that autoconfiguration (both host and "site"--by that I mean local routing scope) is a fundamental idea of IPv6. There are both stateful and stateless methods. > >Third, I agree that summarization is a good idea. But how should it be >implemented? I would think geographically. However, from my personally >experience, that wouldn't work out the best. I've been in a office building >in Minnesota and tracerouted a machine on another floor in the same >building. The path went from Mpls, to Chicago, to St. Louis and back. Any >intelligent summarization will have to be based on the telecommunication >providers rather than geography. Then you have issues of teleco moving, >merging, failing, reconfiguring, etc. I don't know that there is a good >permenent solution. Nobody in the IETF or IRTF has proposed one that met any scrutiny. Geoff Huston has published figures that show the global internet topology is flattening -- becoming less hierarchical -- apparently driven by user multihoming. The demand for multihoming seems driven both by desires for fault tolerance and traffic engineering. Interesting, there is no significant growth in transit AS, only in end user AS. Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=42989&t=42913 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
