RE: CP-->MP Issue (was RE: Polling for new WG I-Ds)

[Dan Li <danli@huawei.com>]

Hi all,

The main motivation for this work is MP->CP migration. Our objective is to 
be able to introduce a control plane into an existing network without 
disrupting user traffic.

We understand that there are operators who would not want to perform this 
function. They might prefer to set up parallel LSPs using the CP and swap 
trafic to them before tearing down the MP LSPs, but we believe that this 
will not be possible in many busy networks. Other operators may prefer to 
have a network outage, but this will not be acceptable for all other operators. 
Therefore, in order to ensure interoperability between different vendors' 
equipment, we believe that a GMPLS process is required for this function. We 
fully expect it to be quite a simple process.

Our main requirement for CP->MP migration is to handle "back-out" of the 
MP->CP migration process. If there is some problem, the operator would like 
to be able to return to how things were before hand (i.e. MP control), 
without disrupting user traffic. We regard this as highly desirable, but 
probably not a mandatory requirement.

Modifying and tearing down LSPs after control plane failure is also a 
possible motivation, but the detailed requirements will need further 
examination. As has been pointed out, tear-down may be achieved relatively 
simply. But modification (for example, re-routing, increase in capacity, 
etc.) may be more complicated.

Other uses of CP->MP migration are interesting, but are not our main 
motivation, but we would like to observe that if the features that meet our 
requirements can also be used for other purposes then that is OK.

Obviously, from the discussions on the mailing list, our I-D does not 
currently present a sufficiently focused description of the motivation for 
this work. We welcome further discussion of the requirements to help us 
improve the draft.

It would be helpful in your discussions if you could understand that we are 
describing the requirements. If it turns out that the requirements can be 
met without changing the protocols, then we are very happy. But if you 
believe that no changes to the protocols are needed, we still need to state 
the requirements. So perhaps our conversation should be about the functional 
requirements not about the solutions.

Best regards,


Dan Li