MIME-Version: 1.0
References: <CAKz0y8yFQdX5w_38rpyvd_sTJsTLxNEXYxxD=ttBzJi=VKxhjg@mail.gmail.com>
 <DB5PR0301MB1909F64D86F4B720BFCCE9679DE50@DB5PR0301MB1909.eurprd03.prod.outlook.com>
 <DB5PR0301MB190988CFD00F53B3BDB95B859DE60@DB5PR0301MB1909.eurprd03.prod.outlook.com>
 <CAA=duU0uBTUSVt3=B5koGmV=hbt0tjVef9uzRRvgqGpQudkwTg@mail.gmail.com>
 <DB5PR0301MB19096B84734126EA9452D5F69DE70@DB5PR0301MB1909.eurprd03.prod.outlook.com>
 <CAAWx_pU3SBvJtcpQRmHyRaS8WHbFmEjRhd3KC4NsghAHV7Qz9Q@mail.gmail.com>
 <CAA=duU12cYHTeYgaMXr9U2jUwaYayuWrBgH+5SHb=NQqk4=TCQ@mail.gmail.com>
 <CAAWx_pUARJ8DER2X1pc60jq7iHT3VJQsitoXAd-gWc9wGvD94Q@mail.gmail.com>
In-Reply-To: <CAAWx_pUARJ8DER2X1pc60jq7iHT3VJQsitoXAd-gWc9wGvD94Q@mail.gmail.com>
From: "Andrew G. Malis" <agmalis@gmail.com>
Date: Tue, 9 Oct 2018 10:54:17 -0400
Message-ID: <CAA=duU1iH2u=ZWs2sZSse6Ntyx_WPUXDUvBt5FWO8nnVjX_Fqg@mail.gmail.com>
To: jwbensley@gmail.com
Cc: Alexander Vainshtein <Alexander.Vainshtein@ecitele.com>, 
 Muthu Arul Mozhi Perumal <muthu.arul@gmail.com>, Shell.Nakash@ecitele.com, 
 Yechiel.Rosengarten@ecitele.com, Michael.Gorokhovsky@ecitele.com, 
 Dmitry.Valdman@ecitele.com, Ron.Sdayoor@ecitele.com, bess@ietf.org, 
 Rotem Cohen <Rotem.Cohen@ecitele.com>
Content-Type: multipart/alternative; boundary="00000000000024718c0577ccea54"
Archived-At: <https://mailarchive.ietf.org/arch/msg/bess/Hr6K5vOzcShgwUiINeE_qi-LpSA>
Subject: Re: [bess] Signaling Control Word in EVPN
Precedence: list

--00000000000024718c0577ccea54
Content-Type: text/plain; charset="UTF-8"

James,

Agreed. We touched on that in section 7 of draft-ietf-pals-ethernet-cw,
where we advised operators that enabling post-CW DPI for ECMP calculations
could cause misordering.

Cheers,
Andy


On Tue, Oct 9, 2018 at 10:35 AM James Bensley <jwbensley@gmail.com> wrote:

> On Tue, 9 Oct 2018 at 15:16, Andrew G. Malis <agmalis@gmail.com> wrote:
> >
> > James,
>
> Hi Andy,
>
> > It's much harder to mandate use of EL than the CW for several reasons:
>
> I didn't say it should be mandated, but recommended.
>
> > - CW implementation is much more common than EL implementation
> > - PWs and/or EVPN are rarely the only traffic in an MPLS traffic tunnel,
> rather, they will be multiplexed with other MPLS-based applications that
> are using the traffic tunnel to reach a common destination. Thus, by using
> the CW, you can disable ECMP only for those MPLS packets that cannot
> tolerate reordering.
>
> The CW does not disable ECMP. Any LSR on the path between ingress and
> egress LER is free to look beyond the MPLS label stack and
> misinterpret the 0x00 0x00 at the start of a control-word as a valid
> MAC that starts 00:00:XX:XX:XX:XX and try to hash on Ethernet headers
> starting directly after the MPLS label stack, and not label stack + 4
> bytes. This is my point. The PWMCW doesn't stop re-ording in all
> cases, but it does in most. So yes, not all devices support EL, but CW
> doesn't stop re-ordering in all cases, so?
>
> > That said, I'm also concerned that because of the existing text in 7432,
> implementations may not be using the CW even for P2P EVPN.
> >
> > And we still don't have a good answer for Muthu's original question. :-)
>
> Sorry my intention is not to send this thread off-topic.
>
> Cheers,
> James.
>

--00000000000024718c0577ccea54
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr"><div dir=3D"ltr">James,<div><br></div><di=
v>Agreed. We touched on that in section 7 of draft-ietf-pals-ethernet-cw, w=
here we advised operators that enabling post-CW DPI for ECMP calculations c=
ould cause misordering.</div><div><br></div><div>Cheers,</div><div>Andy</di=
v><div><br></div></div></div></div><br><div class=3D"gmail_quote"><div dir=
=3D"ltr">On Tue, Oct 9, 2018 at 10:35 AM James Bensley &lt;<a href=3D"mailt=
o:jwbensley@gmail.com">jwbensley@gmail.com</a>&gt; wrote:<br></div><blockqu=
ote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc s=
olid;padding-left:1ex">On Tue, 9 Oct 2018 at 15:16, Andrew G. Malis &lt;<a =
href=3D"mailto:agmalis@gmail.com" target=3D"_blank">agmalis@gmail.com</a>&g=
t; wrote:<br>
&gt;<br>
&gt; James,<br>
<br>
Hi Andy,<br>
<br>
&gt; It&#39;s much harder to mandate use of EL than the CW for several reas=
ons:<br>
<br>
I didn&#39;t say it should be mandated, but recommended.<br>
<br>
&gt; - CW implementation is much more common than EL implementation<br>
&gt; - PWs and/or EVPN are rarely the only traffic in an MPLS traffic tunne=
l, rather, they will be multiplexed with other MPLS-based applications that=
 are using the traffic tunnel to reach a common destination. Thus, by using=
 the CW, you can disable ECMP only for those MPLS packets that cannot toler=
ate reordering.<br>
<br>
The CW does not disable ECMP. Any LSR on the path between ingress and<br>
egress LER is free to look beyond the MPLS label stack and<br>
misinterpret the 0x00 0x00 at the start of a control-word as a valid<br>
MAC that starts 00:00:XX:XX:XX:XX and try to hash on Ethernet headers<br>
starting directly after the MPLS label stack, and not label stack + 4<br>
bytes. This is my point. The PWMCW doesn&#39;t stop re-ording in all<br>
cases, but it does in most. So yes, not all devices support EL, but CW<br>
doesn&#39;t stop re-ordering in all cases, so?<br>
<br>
&gt; That said, I&#39;m also concerned that because of the existing text in=
 7432, implementations may not be using the CW even for P2P EVPN.<br>
&gt;<br>
&gt; And we still don&#39;t have a good answer for Muthu&#39;s original que=
stion. :-)<br>
<br>
Sorry my intention is not to send this thread off-topic.<br>
<br>
Cheers,<br>
James.<br>
</blockquote></div>

--00000000000024718c0577ccea54--