|
Dear Martin, This mail is in reference to the http://tools.ietf.org/html/draft-ietf-mpls-tp-gach-gal-01
draft. I have the following queries:
Suppose
I am using GAL for sending the control/management information and using say
Label 30 all along the LSP to send my data plane information from A <->
Z. Assuming a MIP in between say node B, with the MPLS label swap table
indicating that <ingresslabel=30,egressport=p1,egresslabel=30> and the
G-ACH channel is denoted in B as
<ingresslabel=GAL,egressport=p1,egresslabel=GAL>. Now due to some configuration
error an operator modified the label swap table at B as
<ingresslabel=30,egressport=p1,egresslabel=20> and did not make any error
with G-ACH channel (<ingresslabel=GAL,egressport=p1,egresslabel=GAL>)
remained same. So now the control and management plane are working properly but
actual data plane is down. How does this draft address this?
Suppose
there are 6 nodes in an LSP
A<->B<->C<->D<->E<->Z. Suppose there are two MD
one between C and D (MD#1) and another between B and E (MD#2). Does the MEP at
B generate a G-ACH with GAL which reaches node C which is a MIP and it then
forwards it but then C is also a MEP in MD#2 which again generates a G-ACH with
GAL value. How does D differentiate between the MD#1 and MD#2 G-ACH messages?
How do
you ensure G-ACH channels are bidirectional and hops along the same path?
When
control plane is down, why should the management plane also get affected (as
per present draft control and management are piggybacked within one GAL)? Regards, Jishnu A |