To: Balaji venkat Venkataswami <balajivenkat299@gmail.com>,
 mach.chen@huawei.com, mpls@ietf.org
References: <CAHF4apONg86UYSK9qJN_-bkcw0TcdFXMjgad+gSoxwDmsFbOmA@mail.gmail.com>
From: Stewart Bryant <stewart.bryant@gmail.com>
Message-ID: <88650e7a-5fbb-1a99-dbc0-19650db99146@gmail.com>
Date: Mon, 29 Jan 2018 16:38:47 +0000
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101
 Thunderbird/52.5.2
MIME-Version: 1.0
In-Reply-To: <CAHF4apONg86UYSK9qJN_-bkcw0TcdFXMjgad+gSoxwDmsFbOmA@mail.gmail.com>
Content-Type: multipart/alternative;
 boundary="------------AA00DAE19062554D1451C964"
Content-Language: en-GB
Archived-At: <https://mailarchive.ietf.org/arch/msg/mpls/vMDSJvTNYvSqQybQvbvwSbQ55WQ>
Subject: Re: [mpls] I-D Action: draft-ietf-mpls-sfl-framework-01.txt
Precedence: list

This is a multi-part message in MIME format.
--------------AA00DAE19062554D1451C964
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit



On 29/01/2018 14:50, Balaji venkat Venkataswami wrote:
> Hi Authors,
>
> In the current version of the draft, this follows in section 5.
>
>
>
>     The introduction to an SFL to an existing flow may cause that flow to
>     take a different path through the network under conditions of Equal
>     Cost Multipath (ECMP).  This is turn may invalidate the certain use
>     of the SFL such as performance measurement applications.  Where this
>     is a problem there are two solutions worthy of consideration:
>
>     1.  The operator can elect to always run with the SFL in place in the
>         MPLS label stack.

What this is saying is that you could set up monitoring at for example 
PW or VPN
creation and always run with that extra label. This is not ideal but the
ECMP issue goes away. In other words always run with the SFL in place.



>
>     2.  The operator can elect to use [RFC6790] Entropy Labels in a
>         network that fully supports this type of ECMP.  If this approach
>         is adopted, the intervening MPLS network MUST NOT load balance on
>         any packet field other than the entropy label.  Note that this is
>         stricter than the text in Section 4.2 of [RFC6790].  In networks
>
> Bryant, et al. Expires August 2, 2018 [Page 7]
> Internet-Draft MPLS FI January 2018
>
>         in which the ECMP decision is independent of both the value of
>         any other label in the label stack, and the MPLS payload, the
>         path of the flow with the SFL will be congruent with the path
>         without the SFL.
>

This is an alternative to point 1. Use the EL system (always) and 
require a strict implementation
of it in which the only packet characteristic taken into account when 
selecting the ECMP path
is the value of the EL.

The text seems clear, but if you have any changes in mind please let me 
know.

Best Regards

- Stewart

> Question ?
>
> You dont seem to be considering the consequences of point 1 in the 
> explanation given.
> Or are you trying to say that if point 1 is what is followed, the ECMP 
> decision should be
> independent of the SFL label in place and the MPLS payload, so that 
> the path of the
> flow with the SFL in place will be congruent with the path without the 
> SFL ?
>
> Are you specifying something in point 2 that also applies to point 1 ?
>
> Please clarify.
>
> thanks and regards,
> balaji venkat


--------------AA00DAE19062554D1451C964
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: 8bit

<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <p><br>
    </p>
    <br>
    <div class="moz-cite-prefix">On 29/01/2018 14:50, Balaji venkat
      Venkataswami wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAHF4apONg86UYSK9qJN_-bkcw0TcdFXMjgad+gSoxwDmsFbOmA@mail.gmail.com">
      <div dir="ltr">
        <div>
          <div>
            <pre>Hi Authors,

</pre>
            <pre>In the current version of the draft, this follows in section 5.


</pre>
            <pre>

   The introduction to an SFL to an existing flow may cause that flow to
   take a different path through the network under conditions of Equal
   Cost Multipath (ECMP).  This is turn may invalidate the certain use</pre>
          </div>
        </div>
      </div>
    </blockquote>
    <blockquote type="cite"
cite="mid:CAHF4apONg86UYSK9qJN_-bkcw0TcdFXMjgad+gSoxwDmsFbOmA@mail.gmail.com">
      <div dir="ltr">
        <div>
          <div>
            <pre>
   of the SFL such as performance measurement applications.  Where this
   is a problem there are two solutions worthy of consideration:

   1.  The operator can elect to always run with the SFL in place in the
       MPLS label stack.</pre>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    What this is saying is that you could set up monitoring at for
    example PW or VPN<br>
    creation and always run with that extra label. This is not ideal but
    the<br>
    ECMP issue goes away. In other words always run with the SFL in
    place. <br>
    <br>
    <br>
    <br>
    <blockquote type="cite"
cite="mid:CAHF4apONg86UYSK9qJN_-bkcw0TcdFXMjgad+gSoxwDmsFbOmA@mail.gmail.com">
      <div dir="ltr">
        <div>
          <div>
            <pre>

   2.  The operator can elect to use [RFC6790] Entropy Labels in a
       network that fully supports this type of ECMP.  If this approach
       is adopted, the intervening MPLS network MUST NOT load balance on
       any packet field other than the entropy label.  Note that this is
       stricter than the text in Section 4.2 of [RFC6790].  In networks

<span class="gmail-m_ftr">Bryant, et al.           Expires August 2, 2018                 [Page 7]</span>
<span class="gmail-m_hdr">Internet-Draft                   MPLS FI                    January 2018</span>

       in which the ECMP decision is independent of both the value of
       any other label in the label stack, and the MPLS payload, the
       path of the flow with the SFL will be congruent with the path
       without the SFL.</pre>
            <br>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    This is an alternative to point 1. Use the EL system (always) and
    require a strict implementation<br>
    of it in which the only packet characteristic taken into account
    when selecting the ECMP path<br>
    is the value of the EL.<br>
    <br>
    The text seems clear, but if you have any changes in mind please let
    me know.<br>
    <br>
    Best Regards<br>
    <br>
    - Stewart<br>
    <br>
    <blockquote type="cite"
cite="mid:CAHF4apONg86UYSK9qJN_-bkcw0TcdFXMjgad+gSoxwDmsFbOmA@mail.gmail.com">
      <div dir="ltr">
        <div>Question ?<br>
          <br>
        </div>
        <div>You dont seem to be considering the consequences of point 1
          in the explanation given.<br>
        </div>
        <div>Or are you trying to say that if point 1 is what is
          followed, the ECMP decision should be<br>
        </div>
        <div>independent of the SFL label in place and the MPLS payload,
          so that the path of the <br>
        </div>
        <div>flow with the SFL in place will be congruent with the path
          without the SFL ?<br>
          <br>
        </div>
        <div>Are you specifying something in point 2 that also applies
          to point 1 ?<br>
          <br>
        </div>
        <div>Please clarify.<br>
          <br>
        </div>
        <div>thanks and regards,<br>
        </div>
        <div>balaji venkat<br>
        </div>
      </div>
    </blockquote>
    <br>
  </body>
</html>

--------------AA00DAE19062554D1451C964--