x
<<   <   Page 3 / 3
Sterling Perrin 12/5/2012 | 4:17:32 PM
re: Rumor: T-MPLS Group Gets Shouted Down

This is a great discussion thread (and a great article, Craig), which I'm just catching up on.


Trying to look at this objectively, we have the world's two largest optical equipment suppliers (ALU and Huawei) advocating the bhh variant of OAM&nbsp;- and behind them are at least some of their tier 1 operator customers,&nbsp;who are among the largest operators in the world. How can a standards body find it sensible to ignore a constituency that is this large?? I understand the politics involved, but what industry good is accomplished if these groups are ignored?


On the OAM interoperability issue (between BFD and BHH), is it necessary that they interoperate as standards? I'm asking a question here, not making a point&nbsp;- just trying to understand how critical this piece is, if there are ultimately 2 variants of the standard.


Sterling

Huub_van_Helvoort 12/5/2012 | 4:17:29 PM
re: Rumor: T-MPLS Group Gets Shouted Down

Hello Sterling,


Regarding your question:


"On the OAM interoperability issue (between BFD and BHH), is it necessary that they interoperate as standards? I'm asking a question here, not making a point&nbsp;- just trying to understand how critical this piece is, if there are ultimately 2 variants of the standard."


Interoperability is a corner case, if it exists at all.


Normally operators select one set of tools from the toolbox and use that in their whole network, or in clearly demarkated domains in their network. Connecting domains via other operators/domains will have a client server relation where interoperability is not required.


Between different operators and operator domains the OAM of the link (ETH, OTN, SDH OAM) will be used.


You can compare this with SDH and SONET and how OAM is used in this technology.

Huub_van_Helvoort 12/5/2012 | 4:17:26 PM
re: Rumor: T-MPLS Group Gets Shouted Down

Hello gigEguy, if the interoperation is, as you claim, really an issue, then can you explain why ther will be over 140.000 nodes deployed in networks that use MPLS-TP OAM based on Y.1731 by the end of 2010?


If you read Y.1731 then you will be aware that it can be used in point-2-point and&nbsp; multipoint connections, you will also see that this is a complete toolset that includes OAM for Fault mangement and Performance monitoring and Diagnostics and so will MPLS-TP OAM based on Y.1731.


The MPLS-TP BFD solution that is still under development will support a subset of the Fault mangement of Y.1731 and no Performance monitoring.

gigeguy 12/5/2012 | 4:17:26 PM
re: Rumor: T-MPLS Group Gets Shouted Down

I'm afraid that I must disagree with Huub_van_Helvoort regarding the issue of interoperability. Sure, within a single service provider, that provider is going to pick one OAM solution, which will be reflected in its RFP process. However, if that provider ever wishes to interoperate at the MPLS-TP layer with another provider that selected the other solution, then an MPLS-TP OAM interworking function is going to be needed. That adds cost to the equipment, since the interworking function will have to be defined, implemented, and tested. In addition, just the existance of two solutions makes equipment more expensive, since in reality, all the vendors will need to implement and test both solutions. As was noted earlier in the comments on the article, the Y.1731-based approach is a subset of the functionality of the BFD approach, and is really optimized to handle point-to-point bidiretional co-routed MPLS-TP label switched paths. The BFD-based solution is more general, and can handle unidirectional and point-to-multipoint label switched paths as well.

Huub_van_Helvoort 12/5/2012 | 4:17:25 PM
re: Rumor: T-MPLS Group Gets Shouted Down

Hello nullit, the number I mentioned is indeed from CMCC alone, however other operators are/will be deploying similar numbers of nodes (e.g. CT and CU).&nbsp; (BTW: do you know how many MPLS routers are deployed in the world?)


These nodes are using MPLS-TP OAM based on Y.1731 and are standard compliant, so there is no need to change (I don't want to call it an upgrade) to BFD based MPLS-TP OAM.


Please read draft-fang-mpls-tp-oam-considerations to see that there are also European operators who are in favor of Y.1731 based MPLS-TP OAM. And when you look at the list of contibuting authors of draft-bhh-mpls-tp-oam-y1731 there are even more supporters.


&nbsp;

nullet 12/5/2012 | 4:17:25 PM
re: Rumor: T-MPLS Group Gets Shouted Down

The size of the network you are referring to is probably the CMCC network&nbsp; in China. This is still a single very big network, but it does not indicates that T-MPLS is deployed world wide.


I assume that this network cannot be upgraded to BFD based OAM without major retrofit. This is probably the real reason why the Chinese is so keen to get a solution based&nbsp;on Y.1731 or Draft BHH.


If I am correct, this shows how dangerous it is for operators to deploy pre standard solutions. I would guess that the vendors has promised to be able to upgrade, but has discovered that they are not able to do it at a later stage.


So what is now the future for the BHH draft? What happens if draft BHH does not become an RFC? Will ITU resurrect T-MPLS?

chechaco 12/5/2012 | 4:17:23 PM
re: Rumor: T-MPLS Group Gets Shouted Down

"The BFD-based solution is more general, and can handle unidirectional and point-to-multipoint label switched paths as well."


I would encourage you to demonstrate how BFD is capable to handle uni-directional p2mp. I hope that you're not referring to expired BFD for multi-point networks work that was used in early version of MPLS-TP's CC-CV-RDI work and proved to be an overkill for packet transport network.


It is true, that G.tpoam so far addressed the low hanging fruit in bi-directional co-routed p2p connection. But I would not characterize G.tpoam suit as subset of BFD-based approach at least for the fact that BFD is used only for CC-CV-RDI while LM/DM and LI-LB don't use BFD in any form. At the same time, Y.1731 has strong canonical format that makes HW support so much easier.

menahemk 12/5/2012 | 4:17:00 PM
re: Rumor: T-MPLS Group Gets Shouted Down

Standardization debates reflect commercial interests and this debate is no exception. To a certain extent this is even legitimate. While some actors may pretend to work for the benefit of the mankind, companies fund their standardization activities to gain tangible benefits.


I sense another, less legitimate, trait - IETF aristocracy protecting its positions. Aristocracies occasionally make this mistake of disrespect and disregard for rising powers. One way or another they pay the price further down the road. In this case the revolution will not take a violent form. Rather, it will be irrelevance of the old order through combination of independent standardization in China and draft bhh becoming de-facto standard elsewhere.


Failing to yield or at least share power at the right moment may chop one's head - literally or virtually.

<<   <   Page 3 / 3
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE