headers
Stewart Bryant | 1 Aug 2005 12:10
Picon
Favicon

Re: Demultiplexing of VCCV packets in SAToP and CESoPSN


Yaakov Stein wrote:

> TDMoIP has always had,  and still has,
> its own OAM mechanism, which is similar to VCCV
> and runs over UDP.

Why does it not use VCCV?

> In the recent versions we have moved it to an appendix.
> 
> We indicate the OAM packets by 0001 in the
> CW, just as in the MPLS case.
> 

This has the same problem of how to do the demux in some
fast clean way.

- Stewart
Permalink | Reply | 0 comments |
headers
Yaakov Stein | 1 Aug 2005 13:25
Favicon

RE: Demultiplexing of VCCV packets in SAToP and CESoPSN


> Yaakov Stein wrote:
> 
> > TDMoIP has always had,  and still has, its own OAM 
> mechanism, which is 
> > similar to VCCV and runs over UDP.
> 
> Why does it not use VCCV?

1) The TDMoIP OAM predated VCCV.

2) VCCV does not have the required functionality as yet
(I plan on submitting an ID on the required extensions.)

> 
> > In the recent versions we have moved it to an appendix.
> > 
> > We indicate the OAM packets by 0001 in the CW, just as in the MPLS 
> > case.
> > 
> 
> This has the same problem of how to do the demux in some fast 
> clean way.

There are only problems when RTP is used
(yet another artifact of having RTP stuffed down our throats).

Y(J)S
Permalink | Reply | 1 comment |
headers
Thomas D. Nadeau | 1 Aug 2005 12:33
Picon
Favicon

Re: Demultiplexing of VCCV packets in SAToP and CESoPSN


     Why doesn't it use VCCV which is used by
all other types of PWs?  From an architectural,
operational, and implementation perspective,
it would seem like a better idea to use the
same tool to diagnose/trouble-shoot/verify
all PW types.

     --Tom

> TDMoIP has always had,  and still has,
> its own OAM mechanism, which is similar to VCCV
> and runs over UDP.
> In the recent versions we have moved it to an appendix.
>
> We indicate the OAM packets by 0001 in the
> CW, just as in the MPLS case.
>
> Earlier versions used specific PW labels
> (i.e. UDP port numbers for UDP/IP), but this is problematic
> in the MPLS case since packets in separate LSPs
> may have different latency, loss and PDV characteristics.
>
>
> Y(J)S
>
Permalink | Reply | 0 comments |
headers
Stewart Bryant | 1 Aug 2005 12:35
Picon
Favicon

Re: Demultiplexing of VCCV packets in SAToP and CESoPSN


> 2) VCCV does not have the required functionality as yet
> (I plan on submitting an ID on the required extensions.)

Please do.

> 
> 
>>>In the recent versions we have moved it to an appendix.
>>>
>>>We indicate the OAM packets by 0001 in the CW, just as in the MPLS 
>>>case.
>>>
>>
>>This has the same problem of how to do the demux in some fast 
>>clean way.
> 
> 
> There are only problems when RTP is used
> (yet another artifact of having RTP stuffed down our throats).

Is RTP ever turned on in a deployed network running TDM PWs (of
any flavour)?

- Stewart
Permalink | Reply | 0 comments |
headers
Sasha Vainshtein | 1 Aug 2005 12:37
Favicon

RE: Demultiplexing of VCCV packets in SAToP and CESoPSN

Stewart and all,
Yes, RTP IS used with commerically deployed TDM PWs.

Regards,
				Sasha

> -----Original Message-----
> From: Stewart Bryant [mailto:stbryant <at> cisco.com]
> Sent: Monday, August 01, 2005 12:36 PM
> To: Yaakov Stein
> Cc: Sasha Vainshtein; Skip Booth; Bill Storer (bstorer); 
> pwe3 <at> ietf.org; Tnadeau (E-mail); Alik Shimelmits
> Subject: Re: Demultiplexing of VCCV packets in SAToP and CESoPSN
> 
> 
> 
> > 2) VCCV does not have the required functionality as yet
> > (I plan on submitting an ID on the required extensions.)
> 
> Please do.
> 
> > 
> > 
> >>>In the recent versions we have moved it to an appendix.
> >>>
> >>>We indicate the OAM packets by 0001 in the CW, just as in the MPLS 
> >>>case.
> >>>
> >>
> >>This has the same problem of how to do the demux in some fast
(Continue reading)

Permalink | Reply | 1 comment |
headers
Yaakov Stein | 1 Aug 2005 12:56
Favicon

RE: Demultiplexing of VCCV packets in SAToP and CESoPSN

TDMoIP has always had,  and still has,
its own OAM mechanism, which is similar to VCCV
and runs over UDP.
In the recent versions we have moved it to an appendix.

We indicate the OAM packets by 0001 in the
CW, just as in the MPLS case.

Earlier versions used specific PW labels
(i.e. UDP port numbers for UDP/IP), but this is problematic
in the MPLS case since packets in separate LSPs
may have different latency, loss and PDV characteristics.

Y(J)S
Permalink | Reply | 2 comments |
headers
Thomas Kernen | 1 Aug 2005 13:57
Picon
Favicon

Re: RE: Demultiplexing of VCCV packets in SAToP and CESoPSN


I'll second that.

Regards
Thomas

Sasha Vainshtein wrote:
> Stewart and all,
> Yes, RTP IS used with commerically deployed TDM PWs.
> 
> Regards,
> 				Sasha
> 
>
Permalink | Reply | 0 comments |
headers
Stewart Bryant | 1 Aug 2005 15:36
Picon
Favicon

PWE3 IANA policy - rough consesnus


It is clear that we cannot get the unanimous agreement
of the PWE3 WG on the IANA policy issue.

In order to move forward the chairs need to make a call
on the rough consensus.

Taking into account all of the views that have been
expressed we think that the following is the appropriate
policy:

    0-1024    Expert review
1025-4096    Reserved
4097-        Proprietary - Allocated FCFS by IANA

- Stewart/Danny
Permalink | Reply | 2 comments |
headers
Skip Booth | 1 Aug 2005 16:05
Picon
Favicon

Re: Re: Demultiplexing of VCCV packets in SAToP and CESoPS N


Sasha Vainshtein wrote:
> Skip and all,
> Stitching between MPLS and L2TP PWs is surely an important goal.
> However, I think that the existing situation makes it near impossible - 
> and the TDM PWs are the least complicated part of it!
> 
> When it come to Layer 2 PWs (FR, ATM, HDLC, Ethernet), the L2TP and
> MPLS sequencing schemes differ substantially:
> - L2TP uses a 24-bit no-gap sequence number space
> - MPLS uses a 16-bit, skipped-zero one.

Sure - but there's ways to perform the 16-24 bit conversion, albiet not that
pretty, you can still do it.  In environments where sequencing is not required,
the interworking is pretty straightforward.

For TDM applications, since we are using the PWE3 MPLS CW for both L2TPv3 and
MPLS pseudowires, the interworking is very straightforward assuming we have a
consistent CW/RTP ordering.

> 
> IMHO handling this difference at the stitching point would be much more
> complicated than handling the difference in the CW and RTP order for
> TDM PWs.
> 
> Did I miss something?

Assuming we use the same CW (which I believe we are in agreement on), then the
interworking is very straightforward.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Sasha Vainshtein | 1 Aug 2005 16:07
Favicon

RE: PWE3 IANA policy - rough consesnus

Stewart and all,
I am OK with this.

Regards,
				Sasha

> -----Original Message-----
> From: Stewart Bryant [mailto:stbryant <at> cisco.com]
> Sent: Monday, August 01, 2005 3:36 PM
> To: pwe3
> Cc: W. Mark Townsley; Danny McPherson
> Subject: [PWE3] PWE3 IANA policy - rough consesnus
> 
> 
> 
> It is clear that we cannot get the unanimous agreement
> of the PWE3 WG on the IANA policy issue.
> 
> In order to move forward the chairs need to make a call
> on the rough consensus.
> 
> Taking into account all of the views that have been
> expressed we think that the following is the appropriate
> policy:
> 
>     0-1024    Expert review
> 1025-4096    Reserved
> 4097-        Proprietary - Allocated FCFS by IANA
> 
> - Stewart/Danny
(Continue reading)

Permalink | Reply | 0 comments |

Gmane