[mpls] 答复: MPLS-RT review of draft-rekhter-mpls-pim-sm-over-mldp

[Xuxiaohu <[e-mail]>]

Hi Yakov,

> -----邮件原件-----
> 发件人: Yakov Rekhter [mailto:[e-mail]]
> 发送时间: 2013年9月25日 22:12
> 收件人: Xuxiaohu
> 抄送: Loa Andersson; Lizhong Jin; Gregory Mirsky; Rajiv Asati (rajiva);
> [e-mail]; [e-mail];
> VIGOUREUX, MARTIN (MARTIN); [e-mail]
> 主题: Re: MPLS-RT review of draft-rekhter-mpls-pim-sm-over-mldp
> 
> Xiaohu,
> 
> > Hi,
> >
> > I have reviewed the document. I think this doc is coherent, technically sound
> > and ready to be considered for WG adoption.
> 
> Thanks.
> 
> > Two minor technical comments:
> >
> > 1) "... An implementation MAY support Option 1, as described in Section 2 of
> >    this document. An implementation MUST support Option 2, as described
> >    in Section 3 of this document."
> 
> > Comment: It would be better if the reason for the above implementation
> > suggestion can be given.
> 
> I do not have any strong reasons for the above choice. As I mentioned
> to Loa, I would really like the WG to determine which of the two is
> MUST and which one is MAY.
> 
> > 2) "... This document uses BGP Source Active auto-discovery routes, as
> >    defined in [MVPN-BGP]. This document also identifies the deployment
> >    scenarios where BGP Source Active auto-discovery routes will not be
> >    used."
> >
> > Comment: Besides the approach of originating Source Active auto-discovery
> > routes, shouldn't the MSDP-based approach be considered as well in option
> 1?
> 
> For one thing, use of the MSDP-based approach would require either
> (a) some sort of provisioning to establish a full mesh of MSDP peerings
> among all the edge routers, or (b) some sort of auto-discovery.
> Also, it is far from obvious that such full mesh scale any better
> than use of BGP with Source Active auto-discovery routes.

Thanks for the above explanation. Would it be better to include the above analysis text in the doc for clarification purposes?

Best regards,
Xiaohu

> > Several editorial comments:
> >
> > 1) "This document describes how to map mLDP P2MP tress to multicast
> > trees created by PIM-SM in ASM mode."
> >
> > /s/tress/trees
> 
> Thanks ! Will fix.
> 
> >
> > 2) "An implementation MAY support Option 1, as described in Section 2 of
> >    this document. An implementation MUST support Option 2, as described
> >    in Section 3 of this document."
> >
> > /s/Section2/Section3  /s/Section 3/Section 4
> 
> Sorry, my mistake. Will fix.
> 
> >
> > 3) "...One other scenario where no Source Active auto-discovery routes will b
> e
> >    used is described in section "Originating Source Active auto-
> >    discovery routes".
> >
> > /s/section/section 4.2 (since section 3.1 has the same title as section 4.2)
> 
> Sorry, my mistake. Will fix.
> 
> > 4) Source:  IPv6 RP address, 16 octets.
> > /s/Source/RP
> 
> Sorry, my mistake. Will fix.
> 
> Yakov.