Re: Draft charter for L3VPN: Internet transparency

Mark> Yes, QOS  and SLA are issues,  but folks have  been deploying CE-based
Mark> VPNs on the Internet for  years; evidently they don't consider these a
Mark> show-stopper. 

If a provider  came to you and said,  "for a lot of money,  I will provision
and manage your  enterprise backbone; however, you will not  be able to hold
me  accountable for  the service",  would you  buy it?   Sounds like  a good
business if you can get it!  

Provider-provisioned L3VPNs  are meant to  compete with ATM and  Frame Relay
services backbone  services.  User-provisioned  CE-based VPNS such  as IPsec
tunnels  from  a residence  to  a  corporate  backbone compete  with  dialup
services. 

Mark> Should the charter REQUIRE solutions to be viable across the Internet?
Mark> No.  Should it  encourage development of 1 or  more solutions that are
Mark> viable across the Internet?  Yes.

I wouldn't really  disagree, but I wouldn't want to be  the one developing a
solution which is really viable across the Internet. 

Matt> I'm missing the part  where a VPN has to be used  as the backbone of a
Matt> companies Internet, 

"Intranet."   That's really  the  target  of L3VPN  and  VPWS schemes,  else
you'd just do everything  in a hub and spoke manner.  (I'm  not as sure what
the target of VPLS is.)

Matt> or that QoS and SLAs are an absolute pre-requisite. 

Well, I'd  like to see VPLS  working over arbitrary paths  across the public
Internet, where all you know is you have a best effort datagram service with
no accountability for problems!