The MPLS-OPS Archive

Cell Relay Retreat>MPLS-OPS Archive>month:2003-Jan> msg00055



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Re: About Inter-AS MPLS VPN

  • From: Grzesiek Pawelski <G.Pawelski@elka.pw.edu.pl>
  • Date: Thu, 16 Jan 2003 12:42:47 +0100 (MET)
  • cc: <mpls-ops@mplsrc.com>
  • Resent-Date: Thu, 16 Jan 2003 08:36:09 -0500
  • To: =?EUC-KR?B?udrI8b/1?= <heewoong.park@samsung.com>
  • X-Virus-Scanned: by AMaViS perl-11 mion



On Fri, 10 Jan 2003, [EUC-KR] šÚČńżő wrote:

> The details which I want to get some information are as follows.
> First, the technologies to connect between the ISP/Carriers with MPLS network.
> Second, the cases of current ISPs configuring such connections and their technological methods.
>



You have a few options (according to RFC2547bis):

- BGP exchanging labeled vpnv4 prefixes between ISP ASBRs (MP-eBGP),
  it is suitable especially when you have a bunch of VPNs located
  in the other ISP (one link, simple config, scalable). But the drawback
  is you don't have too much control on the traffic entering your network on
  VPN basis (it is not so easy anyway), and you have to be very careful
  about attributes (esp. RT) carried within the routes you receive,
  as any misconfig or malicious act can mess in your VPNs.

- BGP exchanging labeled vpnv4 prefixes between ISP RRs (multihop
  MP-eBGP), so ASBRs don't need to store or exchange VPN routes, they
  exchange only /32 labeled ipv4 prefixes (MP-eBGP). It is even better
  scalable than the option above.

- vrf back-to-back, ipv4 BGP, you need a separate iface (link) per VPN,
  it can take a lot of resources and config work if you have many VPNs
  exchanged, not scalable. And it is not switched on that hop of course.
  But you have a lot of control then.


Rgds,
Grzesiek

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml