Re: [Bier] MSR6 BOF 1st Issue Category: What is the meaning of “native IPv6"

[Yisong Liu <[e-mail]>]

Hi Greg,

 

Thanks for your question. Please find the response inline. 

 

Best Regards

Yisong Liu

发件人: Greg Mirsky <[e-mail]> 
发送时间: 2022年9月29日 06:17
收件人: Yisong Liu <[e-mail]>
抄送: [e-mail]; [e-mail]; [e-mail]; [e-mail]; BIER WG <[e-mail]>
主题: Re: MSR6 BOF 1st Issue Category: What is the meaning of “native IPv6"

 

Hi Yisong,

thank you for sharing your perspective on multicast technology for IPv6 networks. My understanding of your comparison of BIER with MSR6-TE is that you consider BIER only as applicable in MPLS networks despite BIER WG adopting BIERin6 <https://datatracker.ietf.org/doc/draft-ietf-bier-bierin6/>  that "describes how the existing BIER encapsulation specified in RFC8296 works in a non-MPLS IPv6 network". 

 

Yisong==> Sorry that if the previous expression “Besides the MSR6-TE case, here are the core benefits comparing to the BIER work” causes  any misunderstanding.

In MSR6, there could be 2 types of solution: one is MSR6 BE which could share similar encoding method as BIER but based on IPv6 data plane ( <https://datatracker.ietf.org/doc/draft-lx-msr6-rgb-segment/> https://datatracker.ietf.org/doc/draft-lx-msr6-rgb-segment/ ), other method could also be considered if necessary ( <https://datatracker.ietf.org/doc/draft-chen-pim-be-mrh/> https://datatracker.ietf.org/doc/draft-chen-pim-be-mrh/ ); the other is MSR6 TE which encodes the multicast tree in the packet which could provide the capability of traffic engineering.

If I get this right, maybe your want to compare MSR6 BE with BIERin6 defined in BIER WG?

 

Hence my question: What, in your opinion, is a limitation of the BIERin6 solution that requires the introduction of yet another IPv6 Extension Header, thus adding to the complexity of multicast in the IPv6 network?

 

Yisong==> I think MSR6 BE and BIERin6 are designed for different cases, just like SRv6 and  SR MPLS over UDP.  For example, if the network is based on IPv6, MSR6 BE could be deployed; if the network is based on MPLS with some IPv6 nodes which are supposed to be transited through, BIERin6 may be suitable.

 

Regards,

Greg

 

 

On Wed, Sep 28, 2022 at 1:05 AM Yisong Liu <[e-mail] <mailto:[e-mail]> > wrote:

Hi Joel,

 

Thanks for your response!

To your further question: “Your descriptions here do not explain why using a new routing header is better than using BIER, or any of the other approaches that are being proposed for enhancing multicast handling.  It still requires that the replication devices be enhanced with new forwarding plane capabilities.” Here is some response:

MSR6 is a stateless multicast based on IPv6 data plane by using explicit encoding the destination nodes and optionally the intermediate nodes along the path to these destination nodes in the IPv6 extension header(s). MSR6 is designed for SP or network domain which uses IPv6 rather than MPLS or other data plane.

Besides the MSR6-TE case, here are the core benefits comparing to the BIER work.: 

1．  Allocation and management of IPv6 addresses. 

2．  Simplify the Service identifier by using IPv6 address without further requiring VXLAN/GENEVE

3．  Securing the Service Provider network based on the IPv6 address management mentioned above. 

4．  Reusing IPv6 extension header and the corresponding function, e.g., ESP;

All these benefits coming from building on IPv6 data plane, and re-using the architecture of SRv6. And the benefits have already been discussed and agreed (in some degree especially with the SP who are willing to deploy IPv6) in SRv6 .

 

Best Regards

Yisong Liu

 

发件人: Yisong Liu <[e-mail] <mailto:[e-mail]> > 
发送时间: 2022年9月21日 15:49
收件人: '[e-mail] <mailto:[e-mail]> ' <[e-mail] <mailto:[e-mail]> >
抄送: '[e-mail] <mailto:[e-mail]> ' <[e-mail] <mailto:[e-mail]> >; '[e-mail] <mailto:[e-mail]> ' <[e-mail] <mailto:[e-mail]> >; '[e-mail] <mailto:[e-mail]> ' <[e-mail] <mailto:[e-mail]> >; '[e-mail] <mailto:[e-mail]> ' <[e-mail] <mailto:[e-mail]> >
主题: MSR6 BOF 1st Issue Category: What is the meaning of “native IPv6"

 

Hi all

 

Here are the responses for the 1st Issue Category: What is the meaning of “native IPv6”?, including issue 1-3.

 

 <https://github.com/MSR6-community/MSR6-Issue-List/issues/1> What do you mean by native IPv6? #1 

[Response] We use native IPv6 to describe IPv6 packet running on some media (or data-link layer). E.g., RFC2529 mentions “native IPv6 over most media / ATM” and “IPv6 over IPv4 tunnels” , the latter is treated as opposite concept of “native IPv6”.It is also mentioned in the discussion: “if you are using new forwarding information, this is not native. Putting multicast forwarding information in an IPv6 EH is not native”. IPv6 EH brings extra forwarding behavior, and it is explained in the next response.

 

 <https://github.com/MSR6-community/MSR6-Issue-List/issues/2> What is alternative to native IPv6? IPv6 includes IPv6 EH and SRv6? #2

[Response] As in the answer to issue #1, the alternative to native IPv6 is IPv6 over some kind of tunnel. E.g, IPv6 over IPv4 tunnel, or IPv6 over MPLS tunnel. In our understanding, IPv6 header and IPv6 header with EH, as SRv6, both belong to “native IPv6”, as long as it is not running over some tunnel. E.g., RFC8200 says, “The changes from IPv4 to IPv6 fall primarily into the following categories ... Improved Support for Extensions and Options.”

 

 <https://github.com/MSR6-community/MSR6-Issue-List/issues/3> Don’t like hearing this is called “native IPv6”. Because this also involves a different encapsulation and is not existing IPv6 encapsulation and parse process #3

[Response] Yes, MSR6 also involves encapsulating an original multicast packet into an IPv6 header with an extension header. As the response in the previous 2 questions, we think it is in the scope of “native IPv6”, over no tunnel .If people still have any concern of using “native IPv6”, maybe we could consider to modify the term to for example “ solution based on IPv6 data plane” ?

 

If you have further comments, please let us know.

 

 

Best Regards

Yisong Liu