Bill asks "The stats indicate that the naps don't loose any packets, but folks perceive that the naps are the bottlenecks. What is the situation?"
Steve Feldman says that there have been problems at MAE-WEST. There will also be problems when a customer connects to a shared facility and introduces to much data into the shared port.
Steve Schness echoes the same issues Steve Feldman mentioned as applicable to the Sprint NAP. This was particular true of the shared FDDI. Most of the other problems have been transient in nature.
AADS has not seen any loss.
PAC*BELL has had packet loss in the distant past. The ABR interconnection between the switches and the improved buffer management in the switches has made loss a thing of the past.
Bill compares the various challanges noted by the various providers.
Low speed access may be a serious issue. PAC*BELL is studying it. Sprint notes that at least two providers have stated that they won't peer with a provider that is not connected to at least two NAPS at DS-3.
Chris from ANS asks how the RFC 1490 to RFC 1577 is going. AADS says that there have been some problems, but it appears that no-one has been forced to make the transition. PAC*BELL also sees the same situation. ANS may want to switch to RFC 1577. Steve Feldman notes that the Cisco AIP allows such encapsulation on a per VC basis. Enke Chen from MCI is still using RFC 1490. He asks about OC-3C access. MCI needs it. Steve Feldman says that MFS will offer it on a case-by-case basis. It is not universally available throughout MFS. There is also a limitation of the Gigaswith. PAC*BELL does not want to offer OC-3C available until PAC*BELL is confident that it will work well. However, July is not far away. Sprint is not offering it at this time, but is interested in customer comments. Please contact Steve Schnell about your requirements. However, Sprint is very committed to their FDDI architecture they have in place. Enke Chen is really wanting to know when OC3C AIP interface will be available into the NAP fabric. Steve Feldman does not have a useful way of interconnecting OC-3C into the current infrastructure.
There is a request from Kent England to see AADS report their stats in 5 minutes peaks like everyone else. AADS says that the collector they use is limited on what they can do. Kent asks about the OC-3C connection between the Gigaswitches. The DEC card can't drive it faster than 100Mb/sec.
Bill Norton asks about the staffing of the operation center at the telco NOCs. PAC*BELL comments that their NOC supports fast packet services for PAC*BELL. There are three engineers there that have some knowledge of networking, but there is more work needed. There is also some customer support folks (not the same group) that help resolve customer problems and are knowledgable. MFS has similiar problems, but this is all part of moving the MAE operation from an experiement to a production service. AADS's NOC is being rebuilt at this time and there has been almost 100% turnover. Sprint's NAP is managed by the same engineering group that does IP. The broadband NOC does both ATM and IP. Sprint is considering setting up another NOC that will just do IP (and NAP operations).
Another question about the telco NOCs. Do they have the ability to do IP level debugging? PAC*BELL is going to add this soon. MFS has a management server that can be used by the NOC for IP level testing. They also have a router at MAE-WEST. AADS will consider it, but does not have it currently. Sprint has a router and a host at the NAP for this type of testing.
There is a question about the "web cache". The HARVEST project cache server is available. http://www.nlanr.net. There will be more on this later in this meeting.
Guy Almes has a question for PAC*BELL. Warren's projection seemed to imply exponential growth. It is not exponential. Warren agrees, but exponential smoothing was used since it seems to fit what the curve looked like.
MAE-WEST: The new OC-3C will probably last a couple of months. Another OC-3c link is already planned to handle that.
Are the telco managements responsive to funding more internet efforts?
MFS -- Yes. MFS's chairman actually read the proposed RFC for RSVP PAC*BELL -- Yes. AADS -- There is more support now than there used to be.
What is the situation with Route Servers at the NAPs?
This question will be addressed later in this meeting.
Sprint has no NAP inforamtion page available at this time.