The MPLS-OPS Archive

["McCallum, Robert" <robert.mccallum@thus.net>]

The real problem with VPLS lies not in the technology but the actual SLA that you can offer the customer.  Just ask Cisco the question and all you get back is "we would never recommend a vpls solution due to the SLA issues".  This is the main reason Cisco are very slow to market with VPLS as they
don't actually belive there is a "real" requirement for it. 

Please don't mix this up with L2vpn as this is completely different.  Point to point L2vpns (especially ethernet) are at the moment the hot sell for my company.  In the space of 1 year we have sold hundreds of these types of links throughout the whole of Britain.

If the customer wants any-any connectivity I would always push them down the L3VPN solution.  I would also offer them OSPF, RIPv2 or EIGRP (something which our company differs on other SP's).

HTH

Robert McCallum 
CCIE #8757 R&S
01415663448
07818002241 

> -----Original Message-----
> From: sthaug@nethelp.no [mailto:sthaug@nethelp.no] 
> Sent: 23 March 2004 20:06
> To: luis-m-a-santos@telecom.pt
> Cc: mpls-ops@mplsrc.com
> Subject: RE: [MPLS-OPS]: Question about L2VPN over MPLS
> 
> 
> > What would be the minimum bitrate and what access technology, for 
> > which it would make economic sense to deploy VPLS?  (as opposed to, 
> > for instance, having more "traditional" L3 VPNs such as BGP/MPLS or 
> > Virtual Router)
> 
> I'm still not sure that's the right question. L2 VPNs (VPLS) 
> and L3 VPNs (MPLS/RFC 2547 or similar) serve different 
> purposes. Both are point to multipoint technologies, so they 
> can basically handle much the same topologies. 
> 
> L3VPNs:
> - Have no issues with MAC learning, spanning tree protocol 
> and similar.
> - Nice if you want the ISP do to routing for you.
> - You need to worry about routing protocol redistribution between BGP
> (ISP) and customer routing protocols. Can strongly influence 
> failover times.
> - More mature technology than L2VPNs.
> - IP only.
> 
> L2VPNs:
> - You need to worry about MAC learning, spanning tree 
> protocol and similar. Can strongly influence failover times 
> (particularly with traditional 802.1d).
> - Nice if you want to do the routing yourself.
> - No issues with routing protocol redistribution.
> - Newer technology than L3VPNs, less equipment/fewer vendors 
> that support it.
> - Protocol independent.
> 
> In some ways L2VPNs are "simpler" - but in reality this is 
> only the case if you're familiar with L2 networks and their issues.
> 
> Yes, there are time-sensitive L2 protocols (e.g. LAT, various IBM
> protocols) that may need a higher bandwidth (say more than 
> 256 kbps) to work well over an L2 cloud. However, I would 
> probably worry less about that than the other points in the 
> lists above - remember that people have been running L2 
> protocols over low-speed FR links for many years.
> 
> Steinar Haug, Nethelp consulting, sthaug@nethelp.no
> 
> -------
> The MPLS-OPS Mailing List
> Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
> Archive: http://www.mplsrc.com/mpls-ops_archive.shtml
> 

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