Re: [PWE3] WG Poll on draft-jin-pwe3-cbit-negotiation-03.txt ( comment )

[Luca Martini <[e-mail]>]

On 03/08/11 09:06, [e-mail] wrote:
>
> Luca,
> Thank you for the comments. Please see my reply in line.
>
> Regards
> Lizhong
>
>
> Luca Martini <[e-mail]> wrote on 2011-03-08 03:35:08:
>
> > Sorry for the late response , comments below:
> > Luca
> >
> >
> > On 03/03/11 08:41, [e-mail] wrote:
> > >
> > > Luca,
> > > After discussing with the authors, we decide to add the wait for label
> > > release before sending label request, this wouldn't harm and bring
> > > some secure.
> > >
> > > Then are you OK with the following text?
> > >
> > > When Local PE changes its control word from NOT PREFERRED to PREFERRED
> > > and only if it already received the remote label mapping message with
> > > C-bit=0, additional procedure will be added as follow:
> > > -i Local PE MUST send a label withdraw message to remote PE if it has
> > > previously sent a label mapping, and wait until receiving a label
> > > release from the remote PE.
> > > -ii Local PE MUST send a label request message to remote PE, and wait
> > > until receiving a label mapping message containing the remote PE
> > > configured control word setting.
> > > -iii After receiving remote PE label mapping with control word
> > > setting, Local PE MUST follow procedures defined in [RFC4447] section
> > > 6 when sending it's label mapping message.
> > >
> > Ok the above text is fine.
> > However you need to address the case when the Local PE changes its
> > control word from  PREFERRED to NOT PREFERRED as well.
> [Lizhong] yes, and already addressed in section 3 at the beginning of
> page4 of this draft.
>
> >
> >
> > > Correspondingly, the modified procedure of PE in figure 1 will be as
> > > follows:
> > >
> > > 1. PE2 changes locally configured control word to PREFERRED.
> > > 2. PE2 will then send label withdraw message to PE1.
> > > 3. PE1 MUST send label release in reply to label withdraw message from
> > > PE2.
> > > 4. Upon receipt of Label release message from PE1, PE2 MUST send label
> > > request messages to PE1 although it already received the label mapping
> > > with C-bit=0.
> > > 5. PE1 MUST send label mapping message with C-bit=1 again to PE2
> > > (Note: PE1 MUST send label mapping with locally configured CW
> parameter).
> > > 6. PE2 receives the label mapping from PE1 and updates the remote
> > > label binding information. PE2 MUST wait for PE1 label binding before
> > > sending its label binding with C-bit set, only if it previously had a
> > > label binding with C-bit = 0 from PE1.
> > > 7. PE2 will send label mapping to PE1 with C-bit=1.
> > >
> > No, the above procedure will result in a LDP error and session shutdown.
> > In LDP you  cannot "update" the FEC.  In step 5 PE1 can only send the
> > label mapping message it has previously sent. So PE1 cannot change the
> > C-Bit without first issuing a label withdraw.
> [Lizhong] I don't think such procedure will result in a sesion down
> according to RFC5036. Actually, this is what we want to standardize
> for the PW-FEC processing, to add a way of updating PW-FEC.
you cannot Just "update" a FEC , as it leads to confusion. The receiving
PE will not know how to handle the packet. The label MUST be released ,
and a new label advertised.
> Another way to update PW-FEC is to let PE1 send label withdraw (maybe
> with wrong C-bit) before sending label mapping in step 5, as a reply
> to the label request. But this will lead to additional control message
> load and is not aligned with "receiving label request" behavior in
> A.1.1 in RFC5036, which makes us to abandon it.
>
I do not see how the explanations in appendix A.1.1 in RFC5036 apply to
this problem.

> >
> > > It is to be noted that the above assume that PE1 is configured to
> > > support CW, however in step 5 if PE1 doesn't support CW, PE1 would
> > > send the label mapping message with C-bit =0, this would result in PE2
> > > in step 7 sending a label mapping with C-bit=0 as per [RFC4447] CW
> > > negotiation procedure.
> > >
> > What is needed is some way to send a message to the remote PE that the
> > C-bit negotiation procedure needs to be re-started.
> >
> > Solution 1
> >
> > one way to get this procedure working is to have PE2 release the label
> > mapping previously sent by PE1.
> > If you do this today, most implementations will simply go to a dead
> > state requiring operator intervention.
> > What should happen is that upon receiving the label release , PE1 should
> > withdraw it's label mapping , and reset the c-bit negotiation state
> > machine , and re-advertise the new label mapping restarting the
> > procedure described in rfc4447 from the beginning.
> [Lizhong] This solution has been discussed before on PWE3 mail list
> (http://www.ietf.org/mail-archive/web/pwe3/current/msg11484.html), and
> I also do not prefer this option.
>
No I disagree with Sasha. A label release can be sent anytime , and that
is a local PE decision. Liberal label retention only implies that labels
are usually stored. A label release can be sent for many reasons , like
for example no more local memory available.

>
> >
> > Solution 2
> >
> > Whenever a PE withdraws a FEC mapping , the remote PE resets the C-bit
> > negotiation procedure. If this results in a different C-bit local value
> > then previously advertised , the local PE MUST go though a label
> > withdraw/label release/label advertisement cycle.
> [Lizhong] Solution 2 has already been discussed before on mail list
> (http://www.ietf.org/mail-archive/web/pwe3/current/msg11497.html), and
> also addressed in 00 version of this draft
> (http://tools.ietf.org/html/draft-jin-pwe3-cbit-negotiation-00).
> Unfortunately, the mechanism can not work in MS-PW case.
>
Actually it works fine in the MS-PW case. S-PEs do not care how many
times the label is withdrawn and advertised.
This solution works perfectly fine within all existing LDP rules.

Luca

> >
> >
> > Both solutions above should work within the  LDP rules and are backward
> > compatible. I like solution 2 a bit better.
> >
> > Luca
> >
> > > Thanks
> > > Lizhong
> > >
> > >
> > > ------------send by Luca
> > > Martini-------------------------------------------
> > > Matthew & Andy,
> > >
> > >
> > > Although I support clarifying this procedure , I believe this document
> > > needs some work.
> > > In section 3 for example :
> > >
> > >    When Local PE changes its control word from NOT PREFERRED to
> > >    PREFERRED and only if it already received the remote label mapping
> > >    message with C-bit=0, additional procedure will be added as follow:
> > >
> > >          -i. Local PE sends label withdraw message to the remote if it
> > >              already sends label mapping message, for it has
> changed its
> > >              control word parameter.
> > >
> > >         -ii. Local PE MUST send a label request messages to peer PE to
> > >              get peer's configured control word parameter before
> sending
> > >              new label mapping message to peer PE.
> > >
> > >        -iii. After receiving the new label mapping message from
> peer PE
> > >              and updating the remote label binding information, the
> > >              Local PE should send label mapping to peer PE
> according to
> > >              procedures defined in [RFC4447].
> > >
> > > What does the item i) mean ? the English is not clear.
> > > Why does the PE have to do ii) ?
> > >
> > > All we really need is a statement on how the c-bit negotiation state
> > > machine can be reset if one end has changed "his mind" .
> > >
> > > Can we fix this document before  issuing the -00 WG draft version ?
> > >
> > > Thanks.
> > > Luca
> > >  
> > > --------------------------------------------------------
> > > ZTE Information Security Notice: The information contained in this
> > mail is solely property of the sender's organization. This mail
> > communication is confidential. Recipients named above are obligated
> > to maintain secrecy and are not permitted to disclose the contents
> > of this communication to others.
> > > This email and any files transmitted with it are confidential and
> > intended solely for the use of the individual or entity to whom they
> > are addressed. If you have received this email in error please
> > notify the originator of the message. Any views expressed in this
> > message are those of the individual sender.
> > > This message has been scanned for viruses and Spam by ZTE
> Anti-Spam system.
> >
> >
>
> --------------------------------------------------------
> ZTE Information Security Notice: The information contained in this mail is solely property of the sender's organization. This mail communication is confidential. Recipients named above are obligated to maintain secrecy and are not permitted to disclose the contents of this communication to others.
> This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the originator of the message. Any views expressed in this message are those of the individual sender.
> This message has been scanned for viruses and Spam by ZTE Anti-Spam system.