>Dan West <[EMAIL PROTECTED] wrote,
>
>Just wondering how those NAP xchange points are
>connected physically. VIA....
>
>GIG ETHER?
>100Mb Fast Ether?
>ATM PVCs?
Things are in transition. I'd guess that the largest number of ports
today are FE with a good deal of transition to GE. There are
operational inter-NAP links at 10GE.
ATM is still definitely present. Remember that not all NAPs are in
one room, but are physically distributed. ATM is natural for those,
although metro optical Ethernet will be more and more common.
FDDI has been used historically but are fading out.
>
>Basically, switched and not routed right? I am trying
>to learn about MPLS also. Anyways....
Switched, in a very restricted way -- for example, at the last NANOG,
Paul Vixie (PAIX/MFN) issued a plea for vendors to make switches
available that did NOT run spanning tree or many of the other
features enterprises crave. In the NAP context, VLANs are basically a
simple multiplexing technique.
Not sure I'd immediately associate MPLS with the NAP environment.
Contrary to urban legend, most tier 1 and tier 2 providers do not
interconnect in NAP fabrics, but via private peering. Since one of
the values of MPLS is using it for traffic engineering and SLA
enforcement, I'd hesitate running it through the relative chaos of a
public NAP.
The whole NAP definition is blurring as hosting centers get more NAP-like.
As far as routing vs. switching, the early NAPs switched in the data
plane but peered to a route server to reduce the number of BGP peers.
There's been a trend toward direct peering, although the route server
peerings are retained for statistics collection.
>
>Anybody in this group actually working at a MAJOR
>peer?
Well-- I build things for them.
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
