On Tuesday, February 25, 2014 12:14:16 AM Gert Doering wrote: > Indeed. I deeply distrust "2Mbit/s on ethernet" thingies > and such, though that stuff seems to work surprisingly > well today - *better* than some of the "2Mbit via ATM > VC, with a carrier not being able to tell us which > shaping values to use, thus losing OAM cells on > saturated links"... > > (The same goes for 200Mbit on GE links, of course - > anything where the router doesn't know the actual > end-to-end speed and would need to either rely on "the > modem" behaving "properly" in regards to > queueing/buffering/ policing, or apply hierarchical > shaping on the router, which will usually be not > available in the platform you have)
I think, as you know, this is all hardware-dependent, and sometimes, software does come into play - and to some extent, different vendors may implement things differently. But by and large, modern routing equipment does policing and shaping really well for this to no longer be an issue. One of the things providers and customers need to agree on, though, is whether policing happens at Layer 2 or Layer 3 only. I see this quite a lot in NNI agreements. In our profession, though, IP folk only trust their packet network. If the network is being handed off to someone else for purposes other than IP Transit, there is a huge reluctance to go for anything that is not traveling on DWDM or SONET/SDH. Not sure how we fix this since we are the IP folk :-). Mark.
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
