During the conference today, I alluded to various providers who are expected to have serious memory problems with 16Mbyte routers. One of the European ICM connectees sent me a piece of mail advising me today that: "[the] box here couldn't cope any more and was crashing every couple of hours or so, because it ran out of mem." and "[we] unfortunately had to stick a filter on the inbound stuff from the US, and made it default route. :( Right now it takes what's left of 690, AlterNet and SprintLink, although it looks like [maybe just] ^1800 1239$ at some point". An important R&E network in Europe, therefore, is no longer able to do the kind of routing to the U.S. that they'd like to, and this will adversely affect their backup arrangements and some connectivity to some parts of the commercial Internet. FYI. They do have approval to buy 7000s, however this won't happen before people start complaining. Also, it will get worse, as this happens to more providers around the world. Aggregate and CIDRize, please! As predicted, global disconnectivity has started happening, and it started today. Sean. - -- Sean Doran <smd@icp.net> Sprint/NSFNET International Connectivity Project
participants (1)
-
Sean Doran