Re: [Detnet] Regarding the model for Active OAM packet

[Mach Chen <[e-mail]>]

Hi Loa,

New ACH type is fine, this is also needed for DetNet OAM, IMHO.  The crucial here is how to "cheat" the replication/elimination nodes that an OAM packet is a "normal" DetNet packet, then they can replicate/eliminate the OAM packets as normal DetNet packets. Otherwise, it needs to introduce additional replication/elimination processing for the DetNet OAM packets.

To support this, I suggest to use ACH without GAL (of cause, a new ACH-type) for DetNet OAM (as bellow) and the "reserved" field  carries sequence number information, the "ACH" can be considered as the d-CW by the replication/elimination nodes. 
+----------+
|S-Label  |
+----------+
|ACH        |
+----------+
| Payload|
+----------+

If GAL is used ( the stack as below), additional processing has to be introduced at the replication/elimination nodes, because they have to parse GAL+ACH to decide how to process. That means, the OAM packets will have different replication/elimination process from the normal DetNet packets.

+----------+
|S-Label  |
+----------+
|GAL        |
+----------+
|ACH        |
+----------+
| Payload|
+----------+

Best regards,
Mach 

> -----Original Message-----
> From: Loa Andersson [mailto:[e-mail]]
> Sent: Friday, September 21, 2018 4:49 PM
> To: Mach Chen <[e-mail]>; Greg Mirsky
> <[e-mail]>
> Cc: DetNet WG <[e-mail]>; János Farkas
> <[e-mail]>; [e-mail]
> Subject: Re: [Detnet] Regarding the model for Active OAM packet
> 
> Mach,
> 
> Admittedly I'm not up to speed on DetNet OAM, but .....
> 
> The ACH is specified like this:
> 
>      0                   1                   2                   3
>      0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
>     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>     |0 0 0 1|Version|   Reserved    |         Channel Type          |
>     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> 
> you can define 65k ACH-types, what stops you from defining a ACH channel
> type for DetNet OAM, then define that the structure of the following octets
> in a way that you see fit (like (yes I'm inventing as I type, more thoughts
> should go into to this):
> 
>         0          1          2          3
>         0123 45678901 234567890123 45678901
>        +----+--------+------------+--------+
>        | R  |   LEN  |     relevant info   |
>        +----+--------+------------+--------+
>        |0000|             d-CW             |
>        +----+--------+------------+--------+
>        |        more relevant info         |
>        +----+--------+------------+--------+
> 
> 
> What is that I'm missing?
> 
> /Loa
> 
> On 2018-09-21 14:22, Mach Chen wrote:
> > Hi Loa,
> >
> > Can you clarify how a new ACH-type can address the problem?
> >
> > Best regards,
> > Mach
> >
> >> -----Original Message-----
> >> From: Loa Andersson [mailto:[e-mail]]
> >> Sent: Thursday, September 20, 2018 6:14 PM
> >> To: Mach Chen <[e-mail]>; Greg Mirsky
> >> <[e-mail]>
> >> Cc: DetNet WG <[e-mail]>; János Farkas
> >> <[e-mail]>; [e-mail]
> >> Subject: Re: [Detnet] Regarding the model for Active OAM packet
> >>
> >> Mach,
> >>
> >> I'd like Stewart or Matthew to look at this, but as I understand it
> >> it is possible to define a new ACH-type that can do exactly what you want.
> >>
> >> /Loa
> >>
> >> On 2018-09-20 17:58, Mach Chen wrote:
> >>> Loa,
> >>>
> >>> GAL is just an OAM indicator, the problem here is that when do
> >>> DetNet
> >> OAM, the d-CW will replaced by ACH or by GAL+ACH. No matter which
> way
> >> is used, to support the replication or elimination, there has to be a
> >> sequence number filed. But ACH (as its current defined) does not have
> such a field.
> >>>
> >>> My suggestion is to use the reserved field of ACH to carry sequence
> >> number of OAM packet,  and for those replication or elimination
> >> nodes, they do not have to differentiate whether a packet is OAM
> >> packet or a normal packet, they could just treat the right 28 bits of
> >> the ACH as the sequence number ( or treat the ACH as the d-CW), then
> >> both OAM and replication/elimination can be supported.
> >>>
> >>> Best regards,
> >>> Mach
> >>>
> >>>> -----Original Message-----
> >>>> From: detnet [mailto:[e-mail]] On Behalf Of Loa
> >>>> Andersson
> >>>> Sent: Thursday, September 20, 2018 3:21 PM
> >>>> To: Mach Chen <[e-mail]>; Greg Mirsky
> >>>> <[e-mail]>
> >>>> Cc: DetNet WG <[e-mail]>; János Farkas
> >>>> <[e-mail]>; [e-mail]
> >>>> Subject: Re: [Detnet] Regarding the model for Active OAM packet
> >>>>
> >>>> Mach,
> >>>>
> >>>> If I understand you correctly this is for an LSP in an MPLS
> >>>> network, can you help me understand why GAL does not enough.
> Given
> >>>> that there might be some minor extensions to GAL because of
> >>>> replication and
> >> elimination.
> >>>>
> >>>> /Loa
> >>>>
> >>>> On 2018-09-19 14:31, Mach Chen wrote:
> >>>>> Hi Greg,
> >>>>>
> >>>>> Indeed, there is no DetNet Associated Channel defined in
> >>>>> draft-ietf-detnet-dp-sol-mpls for now, I think there should be.  I
> >>>>> also assume that PW ACH will be used for DetNet OAM.
> >>>>>
> >>>>> Assume that PW ACH will be used for DetNet OAM and the reserved
> >>>>> filed of the PW ACH will be used to carry sequence number for OAM
> >> packet.
> >>>>> But
> >>>>>     for PREF, a tricky way is to treat the “Version”+ “Reserved” +
> >>>>> ”Channel type” as the Sequence number, the replication or
> >>>>> elimination nodes do not need to differentiate whether it is a
> >>>>> d-CW or  a
> >> PW ACH .
> >>>>> This way, OAM can be supported without additional processing and
> >> states.
> >>>>>
> >>>>>           0                   1                   2
> >>>>> 3
> >>>>>
> >>>>>           0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7
> >>>>> 8 9
> >>>>> 0
> >>>>> 1
> >>>>>
> >>>>>
> >>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>>>>
> >>>>>          |0 0 0 1|Verion |    Reserved   |         Channel Type
> >>>>> |
> >>>>>
> >>>>>
> >>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>>>>
> >>>>> Regarding sequence number, there are two ways to generate the
> >>>> sequence
> >>>>> number IMHO:  1) generated by the edge node, but it may need to
> >>>>> configure the start number, or 2) copied from the application-flow
> >>>>> (if there is). If the WG agree with this, the model can be updated
> >>>>> reflect this.
> >>>>>
> >>>>> Best regards,
> >>>>>
> >>>>> Mach
> >>>>>
> >>>>> *From:*Greg Mirsky [mailto:[e-mail]]
> >>>>> *Sent:* Wednesday, September 19, 2018 11:29 AM
> >>>>> *To:* Mach Chen <[e-mail]>
> >>>>> *Cc:* János Farkas <[e-mail]>; DetNet WG
> >>>>> <[e-mail]>; [e-mail]
> >>>>> *Subject:* Re: Regarding the model for Active OAM packet
> >>>>>
> >>>>> Hi Mach,
> >>>>>
> >>>>> thank you for your attention to my comment and the most expedient
> >>>> response.
> >>>>>
> >>>>> I don't find the DetNet Associated Channel defined in
> >>>>> draft-ietf-detnet-dp-sol-mpls and thus I assumed that OAM packets
> >>>>> that follow the data packet encapsulation defined in that draft
> >>>>> use PW ACH as defined in section 5 RFC 4385: True, it includes 8
> >>>>> bits-long Reserved field that may be defined as OAM Sequence
> >>>>> Number but that
> >>>> had
> >>>>> not been discussed. One is certain, existing nodes do not check
> >>>>> the Reserved field. And without a field to hold the sequence
> >>>>> number, PREF will not handle the OAM packets. Another question,
> >>>>> additional processing and amount of state introduced in the fast
> >>>>> path by the fact that OAM's Sequence Number will have different
> >>>>> length and location in d-CW (differentiating cases by the first nibble).
> >>>>>
> >>>>> Now, if we step back from DetnNet in MPLS data plane
> >>>>> encapsulation, why the control-word, as I understand, is
> >>>>> configurable? I think that the Sequence Number is not
> >>>>> configurable, nor the first nibble. What do you think?
> >>>>>
> >>>>> Regards,
> >>>>>
> >>>>> Greg
> >>>>>
> >>>>> On Tue, Sep 18, 2018 at 7:48 PM Mach Chen
> <[e-mail]
> >>>>> <mailto:[e-mail]>> wrote:
> >>>>>
> >>>>>       Hi Greg,
> >>>>>
> >>>>>       The MPLS DetNet header is defined as below:
> >>>>>
> >>>>>       grouping mpls-detnet-header {
> >>>>>            description
> >>>>>                "The MPLS DetNet encapsulation header information.";
> >>>>>            leaf service-label {
> >>>>>              type uint32;
> >>>>>              mandatory true;
> >>>>>              description
> >>>>>                "The service label of the DetNet header.";
> >>>>>            }
> >>>>>            leaf control-word {
> >>>>>              type uint32;
> >>>>>              mandatory true;
> >>>>>              description
> >>>>>                "The control word of the DetNet header.";
> >>>>>            }
> >>>>>          }
> >>>>>
> >>>>>       Although do not consider Active OAM when design the above
> >>>>>       mpls-denet-header,  seems that it can cover Active OAM case as
> well.
> >>>>>       No matter a normal DetNet packet or an Active OAM packet, there
> >>>>>       should be a CW field, just as defined above.
> >>>>>
> >>>>>       For normal DetNet packets, the CW is the d-CW as defined in the
> >>>>>       draft-ietf-detnet-dp-sol-mpls.
> >>>>>
> >>>>>       For OAM packets, the CW is the "DetNet Associated Channel".
> >>>>>
> >>>>>       Best regards,
> >>>>>       Mach
> >>>>>
> >>>>>
> >>>>>       > -----Original Message-----
> >>>>>       > From: detnet [mailto:[e-mail]
> >>>>> <mailto:detnet-
> >>>> [e-mail]>] On Behalf
> >>>>>       Of Greg Mirsky
> >>>>>       > Sent: Wednesday, September 19, 2018 3:17 AM
> >>>>>       > To: János Farkas <[e-mail]
> >>>> <mailto:[e-mail]>>
> >>>>>       > Cc: DetNet WG <[e-mail] <mailto:[e-mail]>>;
> >>>>> detnet-
> >>>> [e-mail]
> >>>>>       <mailto:[e-mail]>
> >>>>>       > Subject: Re: [Detnet] WG adoption poll
> >>>>> draft-geng-detnet-conf-
> >> yang
> >>>>>       >
> >>>>>       > Hi Janos, et. al,
> >>>>>       > the mpls-detnet-header container is based on the solution
> >> described in
> >>>>>       > draft-ietf-detnet-dp-sol-mpls. Analysis of active SFC OAM
> >>>>> in the
> >>>> proposed
> >>>>>       > MPLS data plane solution in draft-mirsky-detnet-oam points
> >>>>> to the
> >>>> potential
> >>>>>       > problem as result the fact that OAM packet doesn't include
> >>>>> d-CW. I
> >>>> believe
> >>>>>       > that this question should be discussed and, if we agree on
> >>>>> the
> >> problem
> >>>>>       > statement, properly resolved. Until then, I do not support
> >>>>> the adoption
> >>>> of
> >>>>>       > the model that may not be capable to support active OAM.
> >>>>>       >
> >>>>>       > Regards,
> >>>>>       > Greg
> >>>>>       > On Tue, Sep 18, 2018 at 12:03 PM Janos Farkas
> >>>> <[e-mail] <mailto:[e-mail]>>
> >>>>>       > wrote:
> >>>>>       > >
> >>>>>       > > Dear all,
> >>>>>       > >
> >>>>>       > > This is start of a two week poll on making
> >>>>>       > > draft-geng-detnet-conf-yang-04 a working group document.
> >>>>> Please
> >>>> send
> >>>>>       > > email to the list indicating "yes/support" or "no/do not
> >> support".  If
> >>>>>       > > indicating no, please state your reservations with the
> >> document.  If
> >>>>>       > > yes, please also feel free to provide comments you'd like to see
> >>>>>       > > addressed once the document is a WG document.
> >>>>>       > >
> >>>>>       > > The poll ends Oct 3.
> >>>>>       > >
> >>>>>       > > Thanks,
> >>>>>       > > János and Lou
> >>>>>       > >
> >>>>>       > > _______________________________________________
> >>>>>       > > detnet mailing list
> >>>>>       > > [e-mail] <mailto:[e-mail]>
> >>>>>       > > https://www.ietf.org/mailman/listinfo/detnet
> >>>>>       >
> >>>>>       > _______________________________________________
> >>>>>       > detnet mailing list
> >>>>>       > [e-mail] <mailto:[e-mail]>
> >>>>>       > https://www.ietf.org/mailman/listinfo/detnet
> >>>>>
> >>>>>
> >>>>>
> >>>>> _______________________________________________
> >>>>> detnet mailing list
> >>>>> [e-mail]
> >>>>> https://www.ietf.org/mailman/listinfo/detnet
> >>>>>
> >>>>
> >>>> --
> >>>>
> >>>>
> >>>> Loa Andersson                        email: [e-mail]
> >>>> Senior MPLS Expert
> >>>> Bronze Dragon Consulting             phone: [phone]
> >>>>
> >>>> _______________________________________________
> >>>> detnet mailing list
> >>>> [e-mail]
> >>>> https://www.ietf.org/mailman/listinfo/detnet
> >>> _______________________________________________
> >>> detnet mailing list
> >>> [e-mail]
> >>> https://www.ietf.org/mailman/listinfo/detnet
> >>>
> >>
> >> --
> >>
> >>
> >> Loa Andersson                        email: [e-mail]
> >> Senior MPLS Expert
> >> Bronze Dragon Consulting             phone: [phone]
> 
> --
> 
> 
> Loa Andersson                        email: [e-mail]
> Senior MPLS Expert
> Bronze Dragon Consulting             phone: [phone]