Bill remarks that the old NSFNET had a large amount of information available. The new environment seems to have less information available. Is that level of information necessary?
ANS says that such maps are difficult to maintain and in fact ANS does not have a current map. Having the customers know about the backbone map is probably useful, but the level of detail that was available in the NSFNET will probably not be available.
Peter remarks that the internal configuration is probably not important to most users. Others remark this group is not the average group of users. However, Peter insists that having this information may make the situation worse. Randy says that he wants to be able to point out the problems.
Sprint: The snapshot indicated that the MAE-E link was down about .9% of the time which is about 13 minutes per day. Jim could not comment directly on that.
MCI: What about vBNS and the problems with upgrade OC3c and OC12? ___ said that the two groups are not the same and while the experiences are shared, the concerns may not be.
Yakov asks at what level folks want to see this information? IP, ATM VCs or SONET trunks or what? One participant comments that having information about the link level provider as well the ip level information.
General question: The NAP folks say they don't have any loss, so who does?
ANS: There are no guarantees. There is an on-going battle with the RS6000s and some traffic is being moved from AS690 to the AS1673.
MCI: They have not seen loss since the OC-3c upgrade. The interconnects are still overloaded.
Sprint: Backbone links are pretty good. The Interconnect links are heavily loaded. Not the interconnects, but the links themselves.
MCI: MCI is using Cisco hardware in their POPs. There are no plans going to using NET*STAR.
ANS: The IRR would be a good resource of information on how things are connected.
SPRINT: Sprint does not register aut-num objects. Sean says that it's work. Sprint does register some information.
ANS: The Sprint NAP's NSS router will be retired with a 7500 sometime soon.
Guy Almes notes that in the NSFNET days, there was available alot of flow information. ANS: Tinigrams are not liked by the NSS. Moving to the new network has been a priority for those sites that generate lots of tinigrams. MCI and Sprint have not done alot of this type of analysis.