Re: [bess] 答复: 答复: 答复: Flowspec for L2VPN and E-VPN

[Mach Chen <[e-mail]>]

Hi Weiguo, Wim and others,

IMHO, AFI/SAFI based Flowspec would have better scalability and compatibility. There is a precedent (RT-Constrain) that adopted the unified RT for all AFI/SAFI that bring many limitation when deploying RTC.

Best regards,
Mach

> -----Original Message-----
> From: BESS [mailto:[e-mail]] On Behalf Of Haoweiguo
> Sent: Friday, November 14, 2014 8:42 AM
> To: Henderickx, Wim (Wim); Thomas Morin; BESS
> Cc: IDR Chairs
> Subject: [bess] 答复: 答复: 答复: Flowspec for L2VPN and E-VPN
> 
> Hi Wim,
> It seems to be a solution. Another problem:
> Current BGP flow spec for L2 VPN /L3 VPN relies on Rout Target for policy
> import/export. If using unified solution, RT can't overlap between different
> applications(L2VPN,L3VPN...). If using separating AFI/SAFI solution, no RT
> constraint issue.
> Maybe there are other questions for unified solution, i would like to hear other
> expert's comments on your proposal.
> Thanks
> weiguo
> 
> ________________________________________
> 发件人: BESS [[e-mail]] 代表 Henderickx, Wim (Wim)
> [[e-mail]]
> 发送时间: 2014年11月14日 8:27
> 收件人: Haoweiguo; Thomas Morin; BESS
> 抄送: IDR Chairs
> 主题: Re: [bess] 答复:  答复:  Flowspec for L2VPN and E-VPN
> 
> We define a new AFI/SAFI that accommodates all we have + include L2
> extensions.
> Operators that don’t need L2 extensions keep what they have.
> Operators that need L2 extensions go to the new method or mix the new
> method with the old methods per service type.
> 
> Make sense?
> 
> On 13/11/14 14:16, "Haoweiguo" <[e-mail]> wrote:
> 
> >How to achieve compatability with current existed flowspec[RFC5575]
> >applications?
> >Thanks
> >weiguo
> >
> >________________________________________
> >发件人: Henderickx, Wim (Wim) [[e-mail]]
> >发送时间: 2014年11月14日 8:14
> >收件人: Haoweiguo; Thomas Morin; BESS
> >抄送: IDR Chairs
> >主题: Re: 答复: [bess] Flowspec for L2VPN and E-VPN
> >
> >If we define a new things I prefer to address the wider issue and
> >include
> >L2 in that.
> >
> >On 13/11/14 14:13, "Haoweiguo" <[e-mail]> wrote:
> >
> >>Hi Wim,
> >>Allocating different AFI/SAFI(s) for each flow spec application is a
> >>applicable solution. Theoretically, unified mechanism for all flowspec
> >>can be designed, but it maybe a more harder work in IDR.
> >>Thanks
> >>weiguo
> >>
> >>________________________________________
> >>发件人: BESS [[e-mail]] 代表 Henderickx, Wim (Wim)
> >>[[e-mail]]
> >>发送时间: 2014年11月14日 7:55
> >>收件人: Thomas Morin; BESS
> >>抄送: IDR Chairs
> >>主题: Re: [bess] Flowspec for L2VPN and E-VPN
> >>
> >>As I stated in the IDR meeting my observation is that we require to
> >>many
> >>AFI/SAFI(s) for all flow spec functions. Flow spec in general is
> >>providing match criteria¹s with related actions. Given the proposal on
> >>Flowspec for
> >>L2 is new we should look at the bigger picture.
> >>In My view we need a mechanism in BGP to advertise Flowspec match
> >>criteria¹s with related actions and they should cover L2/L3-IPv4/IPv6.
> >>
> >>On 13/11/14 13:44, "Thomas Morin" <[e-mail]> wrote:
> >>
> >>>Hi WG,
> >>>
> >>>A heads up...
> >>>
> >>>These two drafts relate to BESS and thus may be of interest to us:
> >>>- draft-hao-idr-flowspec-l2vpn
> >>><http://tools.ietf.org/html?draft=draft-hao-idr-flowspec-l2vpn-01>
> >>>(on idr agenda, being presented right now)
> >>>- draft-hao-idr-flowspec-evpn
> >>><https://tools.ietf.org/html/draft-hao-idr-flowspec-evpn-00>
> >>>
> >>>Best,
> >>>
> >>>-Thomas
> >>>
> >>>
> >>>_______________________________________________
> >>>BESS mailing list
> >>>[e-mail]
> >>>https://www.ietf.org/mailman/listinfo/bess
> >>
> >>_______________________________________________
> >>BESS mailing list
> >>[e-mail]
> >>https://www.ietf.org/mailman/listinfo/bess
> 
> _______________________________________________
> BESS mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/bess
> _______________________________________________
> BESS mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/bess