The MPLS WG Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Doubt about RFC 3107
Li
AFAIK , it i simply mean that if U have
A-B-C-D
and "A" and "D" are having a label distribution session (using BGP or
Targeted LDP) ,if
"D" distribute label "L1" to "A" so "A" can not use "L1" unless their is an
LSP from A to D .
ie: we need a tunnel (lsp) from A to D .
The catch that "A" have no mean to find out that such LSP exist if
"independent" label
assignement are used .
Ex: "D" advertise a label "L5" to "C" for it's loopback address .
"C" have been configured (say by mistake) to not advertise label for
"D" loopback address or
simply link B-C is not mpls link .
"B" will advertise label "L10" for FEC "D loopback address" to "A" .
"D" advertise to "A" (through BGP or targeted LDP) label "L1" say
for fec 10./8
Now "A" is under the impression that an LSP exist toward D and if it
want to transmit pavket to
10./8 it will create a label stack with {L10,L1} and send to "B" .
B pop the top label ("L10" ,with cisco terminology the action is
Untag) and expect to see plain
IP packet but it is not the case so it will drop the packet .
If the label assignement is "ordered" such pblm do not exist . ie"B" will
never advertised a label
for "D" loopback address unless "B" recieved a label for same .
Brgds
>From: lidefeng <lidefeng@huawei.com>
>To: mpls@UU.NET
>Subject: Doubt about RFC 3107
>Date: Mon, 19 May 2003 17:05:33 +0800
>
>It is stated in section 6 of RFC 3107 that "This is true any time labels
>are
>distributed between non-adjacent LSRs, whether that distribution is done by
>BGP or by some othermethod."
>
>What does it mean?Maybe it is totally wrong at all?
>
>Who can explain it?
>
>Regards
>
>Li Defeng
>
>Huawei technologies
>
>
_________________________________________________________________
STOP MORE SPAM with the new MSN 8 and get 2 months FREE*
http://join.msn.com/?page=features/junkmail
|
|