[mpls] 答复: [Bier] Encapsulation first nibble

> -----邮件原件-----
> 发件人: mpls [mailto:mpls-bounces@ietf.org] 代表 Xuxiaohu
> 发送时间: 2015年3月18日 22:20
> 收件人: Jeffrey (Zhaohui) Zhang; Antoni Przygienda; stbryant@cisco.com; Eric
> Rosen; BIER; IJsbrand Wijnands
> 抄送: mpls@ietf.org; sfc@ietf.org
> 主题: Re: [mpls] [Bier] Encapsulation first nibble
> 
> Hi Jeffrey,
> 
> > -----邮件原件-----
> > 发件人: Jeffrey (Zhaohui) Zhang [mailto:zzhang@juniper.net]
> > 发送时间: 2015年3月18日 21:53
> > 收件人: Xuxiaohu; Antoni Przygienda; stbryant@cisco.com; Eric Rosen;
> > BIER; IJsbrand Wijnands
> > 抄送: mpls@ietf.org; sfc@ietf.org
> > 主题: RE: [Bier] Encapsulation first nibble
> >
> > Xiaohu,
> >
> > > Following such logic (i.e., assign a nibble value specific to the
> > > new encapsulation header), we would have to consider how many
> > > available nibble values are there which can be used for future
> > > encapsulation headers in the same way.
> >
> > There are still quite a few values left; defining a new encapsulation
> > is not lightly taken; adding a new IP version is not lightly taken
> > either. Besides, there is this "reserved value 15" idea from Ice.
> 
> Since the control word is optional when transporting Ethernet over MPLS, I
> wonder whether the idea of reserving the first nibble value of 15 is feasible in
> practice.
> 
> > Coupled with the reasons that Eric gave:
> >
> > > I think the proper direction is to use the MPLS entropy label,
> > > rather than to have each router compute the entropy based on an
> > > analysis of the next encapsulation header.
> > >
> > > So I don't think BIER justifies the use of a new MPLS special
> > > purpose payload-protocol-identifier label.
> >
> > I would think the encap-specific nibble is quite reasonable, and
> > better than a special purpose label.
> 
> Due to the same reason as mentioned above, I wonder whether it's feasible to
> reserve a certain first nibble value as an encap-specific nibble in practice.

More specifically, since it allows the Ethernet frames to be transported over MPLS networks w/o control word, any possible value of the first nibble may have been used. Of course, if the first nibble is immediately after a certain SPL or ESPL, it becomes possible to reserve a certain value of that nibble as a given encapsulation header (see https://tools.ietf.org/html/draft-guichard-sfc-metadata-header-00#page-4). However, in this case (i.e., an SPL or ESPL is used), it seems more straightforward to add a proto field after the first nibble, rather than using the first nibble itself as a poor-man's proto field.

> Best regards,
> Xiaohu
> 
> > Jeffrey
> _______________________________________________
> mpls mailing list
> mpls@ietf.org
> https://www.ietf.org/mailman/listinfo/mpls