The MPLS-OPS Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Re: Fwd: MPLS DiffServ over Ethernet (L-LSP vs E-LSP)
At 10:27 AM 10/19/2002 -0400, Irwin Lazar wrote: The reason for the doubt is: >>The Ethernet header (like ATM) doesn't have any QoS >>information field (unless 802.1q is supported). So >>does that imply that E-LSP can't e used when ethernet >>(without 802.1q support) is the underlying media. Is >>L-LSP the only choic then? You assume that the Ethernet switch carrying the packets is unaware of MPLS. I have yet to see a service provider network design where that is the case. This means that Ethernet might be used to encapsulate the MPLS frame between switches, but at each switch the Ethernet header is stripped off and the queueing/switching performed on the MPLS frame, including the EXP bits. So, to answer your question, using Ethernet as a point-to-point MPLS link technology does not preclude using E-LSPs or L-LSPs. If you are assuming that there is an Ethernet cloud between MPLS switches, then yes, you need to do EXP<->802.1q mapping (and also configure each Ethernet switch with the correct per hop QoS behavior). >> >From the RFCs and drafts on this topic, it appears >>that both kind of LSPs have pros and cons. Has someone >>really implemented L-LSPs ? I've seen implementations of both L-LSP and E-LSP over Ethernet -- I'll let the vendors themselves respond. Regards, Mathew ------- The MPLS-OPS Mailing List Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml Archive: http://www.mplsrc.com/mpls-ops_archive.shtml
|
|