Joe Touch | 1 Dec 2004 01:29
Picon
Favicon

Re: multiaddressing


Wesley Eddy wrote:
> On Tue, Nov 30, 2004 at 12:32:34PM -0800, Joe Touch wrote:
> 
>>It would be useful to review the docs of the shim subgroup of multi6. 
>>They're just now being submitted to be WG docs.
>>
> 
> 
> Specifically, draft-nordmark-multi6dt-shim-00.txt lists one of the shim's
> goals as:
> """
>     o Have no impact on upper layer protocols in general and on
>       transport protocols in particular.
> """
> 
> -Wes

No impact means no detrimental impact. It'd be useless if there's no 
benefit. Granted, they're not working on using multiple connections at 
once, but they are doing all the rest of what was mentioned - including 
shifting addresses as needed and potentially informing a transport or 
application that wants to know (AFAIR). I.e., adding muxing there seems 
the obvious place, rather than at the transport or the application.

Joe
_______________________________________________
tsvwg mailing list
(Continue reading)

giuseppe | 1 Dec 2004 02:25
Picon
Favicon

Add/Delete under ns

Somebody knows whether are some extensions to ns for Add/Delete IP 
features available?
G
Joe Touch | 1 Dec 2004 07:02
Picon
Favicon

Re: multiaddressing


Joe Touch wrote:
> 
> 
> Ted Faber wrote:
> ...
> 
>> Why *not* DNS?  It's a distributed replicated 1-to-many database.  It
>> already returns lists of network addresses associated with a string, and
>> there's a well-known API to get at those.  I don't understand the
>> additional dependency created here unless the endpoints were not
>> exchanging DNS names to begin with. 
> 
> 
> When DNS returning multiple RRs it doesn't mean they're on the same 
> host; they are just 'equivalent' in some fuzzy sense, e.g., replica 
> servers of a database, alternate gateways to a VPN, etc.

s/RRs/records in a reply/

I confused the first R with "rotation" (brain cloud). I'm referring to 
rotation of multiple A record replies, ala RFC1794, FWIW.

Joe

> There would need to be a different record to indicate records that go to 
> the same 'box' (i.e., we could call them WH for 'weak host').
> 
> Joe
(Continue reading)

Randall Stewart | 1 Dec 2004 14:44
Picon
Picon

Re: multiaddressing

Ted/All:

First I am removing all the "cc's and to's" I think we are
all on the list and I don't want to cause multiple copies to
be recieved...:-D

Now, I have kept silent on this for a while now trying to
contemplate my reply so that it might be consistent  and
grokable if thats possible :-D

I am not going to touch the mobility aspect of this since
I have never been a supporter of the use of the SCTP-ADDIP
document as a mobility crutch. The idea is interesting.. don't
get me wrong.. but the document was NOT intended for that purpose.
(Note also that if I remember right from my last peruseal of the
  IPR page both Siemens and Cisco have IPR in for use of
  transport protocols in mobility.. DCCP and SCTP.. might
  want to check that page if you are a fan of this concept).

So I will exclude mobility from my discussion of multi-addressing.
Maybe in the future we can do something for mobility.. but I thought
we did that with mobile IP, maybe working at the lower layers does
not always work.. I don't know  :-/

Now, why does SCTP have multiple addresses... (in
priority order)

1) Fault-tolerance - When the design team for SCTP worked
    on SCTP we wanted to be able to get a message from sender
    to receiver with as much fault tolerance as we could. This is
(Continue reading)

Henderson, Thomas R | 2 Dec 2004 07:57
Picon
Favicon

RE: multiaddressing

Randall, I agree with you on many points of your argument,
but I think the problem is more general than you stated and
will require multiple solutions.  In particular,

> Now, why does SCTP have multiple addresses... (in
> priority order)
> 
> 1) Fault-tolerance - 
> 2) Load-balancing - 

Fault tolerance is an interesting case, because (as it is defined
in SCTP) it means that certain transport PDUs have different
priority or delivery semantics than others.  In particular, 
you want to force timeout retransmissions onto a different path. 
I agree here that if you put this kind of function in a multihoming
shim layer, it makes the interface more complicated to transport--
in this case, you want transport to be able to specify which TPDU
goes where, and to know where TPDUs came from, etc.  So I would
agree that SCTP, being oriented towards signaling transport, has
particular signaling fault tolerance requirements in mind (the
requirements may extend beyond signaling and be generally 
applicable), and it makes sense to try to solve it in SCTP.

For load-balancing, it is not as clear to me-- it could be in
a shim, it could be in transport, it could be in session layer.  
It seems to make sense to integrate it with congestion control,
which might lean it towards transport again, but if it is not
in transport, then I don't believe that the interface is necessarily
that complex.

(Continue reading)

Randall Stewart | 2 Dec 2004 12:42
Picon
Picon

Re: multiaddressing

Henderson, Thomas R wrote:
> Randall, I agree with you on many points of your argument,
> but I think the problem is more general than you stated and
> will require multiple solutions.  In particular,
> 
> 
>>Now, why does SCTP have multiple addresses... (in
>>priority order)
>>
>>1) Fault-tolerance - 
>>2) Load-balancing - 
> 
> 
> Fault tolerance is an interesting case, because (as it is defined
> in SCTP) it means that certain transport PDUs have different
> priority or delivery semantics than others.  In particular, 
> you want to force timeout retransmissions onto a different path. 
> I agree here that if you put this kind of function in a multihoming
> shim layer, it makes the interface more complicated to transport--
> in this case, you want transport to be able to specify which TPDU
> goes where, and to know where TPDUs came from, etc.  So I would
> agree that SCTP, being oriented towards signaling transport, has
> particular signaling fault tolerance requirements in mind (the
> requirements may extend beyond signaling and be generally 
> applicable), and it makes sense to try to solve it in SCTP.

Two comments to the above point. A) SCTP has a broader
scope then its initial signalling world. and B) I am a bit
unclear on how one gets differetn priority's. If you do the
currently recommend approach, retransmissons at timeout goes
(Continue reading)

Kwok Ho Chan | 6 Dec 2004 16:28

Re: Re: Re: some problems aboutdraft-baker-diffserv-basic -classes-0 4

Bei-Bei:
Please notice I have added this thread to the TSVWG mailing list.
Please see response embedded below.
Any additional input are welcomed.
Thanks!
-- Kwok Ho --

At 10:39 PM 12/4/2004, =?gb2312?B?sszd7d3t?= wrote:
Kwok Ho:
 
Thank you very much for your so detailed answer.
Your help have resolved some questions I had.
Of course,you can put this conversation back onto the TSVWG mailing list.:)
 
And,may I understand the draft as follow : the gradation of traffic priority from high to low is
 CS7,CS6.EF,CS5.AF4,CS4,AF3,CS3,AF2,CS2,AF1,CS0(DF),CS1?

I think you can say that.  But in reality, IMHO, the network needs to make sure it
upholds the promise it gave to its customer, no matter which service class it is.
This may mean the network needs to admit less traffic that goes into service class using
EF so the service classes using CS4 and AF4x will still uphold the promise to its users.
The decision is on which service classes a network supports (they don't need to support
all the service classes) and how much of each.

I have cleared about those service classes except for the Broadcast Video Service Class.
Based on the draft,Broadcast video is inelastic traffic,and Multimedia Streaming is elastic.
Why do the former use CS3 and the latter use AF3?
And in 4.2.1 part of draft-baker-diffserv-basic-classes-02, the Multimedia Streaming class include
the Broadcast video service.What is the reason for separating the broadcast video from the multimedia streaming
in draft-baker-diffserv-basic-classes-03(and 04) ?

Broadcast Video service class is meant for High Definition broadcasts and the likes.
We have found that the Loss Tolerance for these traffic is relatively lower than Multimedia
Streaming service class.  Hence its separation.  Again this is based on what the application
requires of the network.  At low loss high capacity parts of the network, both of these
service classes may get treated together without impact to the promises made to the
user of each service class (discussion for the service class aggregation draft). 

 
Thank you very much!
 
 
 
                                                                  Cai bei-bei
ps:Best wishes to you!
¡¡¡¡
 
 
======== 2004-12-02 23:49:52 ÄúÔÚÀ´ÐÅÖÐдµÀ£º ========
 
Bei-Bei:
Agree that they all need low delay, low jitter.
Other factors that affects the service classes placement are loss tolerance, traffic elasticity,
and packet size (and its variation).

Telephony uses EF. the Multimedia Conferencing uses AF because it is elastic traffic.
There is also the Real-Time Interactive class using CS4 that handles inelastic video traffic.
Please reference Figure 1 of
  http://www.ietf.org/internet-drafts/draft-baker-diffserv-basic-classes-04.txt

EF PHB normally get deployed with some kind of upper limit on the traffic it treats, either
with some policer, rate limiter, and/or some other admission control methods.
With this in mind, the EF traffic should not starve the AF4x and CS4 traffic.  Each should
forward the traffic as configured (one may not want to have too much EF traffic that will
add too much delay to the CS4 and AF4x traffic making them un-acceptable).  Without
having the elastic traffic adversely affecting the inelastic traffic.  Or having large packets
adding too much jitter to small inelastic traffic.

This draft takes the points of view of how each application using the different service classes
will receive the network treatment it requires (end to end), when traffic in multiple service classes
need to share the network.  When these traffic traverse both low and high capacity networks/links.

There is another draft that discusses how in high capacity networks/links, multiple service classes
can be aggregated together.  I have published -00 version of this draft but I am in the process of
updating this draft to add more content and making changes based on the comments received.
I would advise to wait for the -01 of draft-chan-tsvwg-diffserv-class-aggr.

I hope this help to answer/resolve the questions/issues you have.

Also, can I put this conversation back onto the TSVWG mailing list?

Thank you!
-- Kwok Ho --


At 08:17 AM 12/2/2004, =?gb2312?B?sszd7d3t?= wrote:
Kwok Ho Chan:
 
Thank you for your answers.
Yes,I am a student of Nanjing University of Posts and Telecommunications of China.
Now we are doing a project about end-to-end IP QoS.
 
Every service has different QoS requirement.So it will be classed, and deployed with DSCP, then be treated differently based on the different class.
Voice and Vedio require real-time, very low delay, very low jitter,which must use EF PHB.So I am
puzzled that Multimedia Conferencing use AF PHB. Services have lower priority ,they may wait in a queue for a longer time, which will bring higher delay. The same question exist between multimedia  streaming and broadcast video and so on.
 
We want to use the draft to class services, and deal with these based on the DSCP. 
 
¡¡¡¡Thank you very much!!
 
======== 2004-11-24 22:54:53 ÄúÔÚÀ´ÐÅÖÐдµÀ£º ========
 
Bei-Bei:
Thank you for your comments.
I think you mis-understood the concepts.
The basic-classes draft separates the traffic into different classes
based on their traffic characteristics.
It does not necessarily mean Multimedia Conferencing is lower than Telephony,
they just needs to be separated.
We also recommend some treatment approaches for each Service Class based
on each of their traffic characteristics.
Please look at (read) the draft this way and let us know if think differently.
Thanks!
-- Kwok Ho Chan --
p.s. Are you based in China?  It looks like a Chinese name to me.
      How may you want to use this draft?


At 09:18 AM 11/24/2004, =?gb2312?B?sszd7d3t?= wrote:

Hi:

  I have some problems about draft-baker-diffserv-basic-classes-04.

  1.Why is the priority of Multimedia Conferencing(eg.H.323/V2 Video conferencing) lower than telephone? The former(video) sometimes need very low delay,eg.CBR video.

  2 Usually,the signalling precede the corresponding service. Why does the service class signalling have a lower priority than telephone? And i also don't know why the QAM is behind Low Latency  data,and the broadcast video is behind multimedia streaming?In my concepte,broadcast video is include in multimeida streaming and need less delay.

 
 
  Thank you very much!!
       

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Cai Bei-bei
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cbeibei <at> 263.net
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2004-11-24

 

= = = = = = = = = = = = = = = = = = = = = =

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡ÖÂ
Àñ£¡
 
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡²ÌÝíÝí
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cbeibei <at> 263.net
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2004-12-02
 

= = = = = = = = = = = = = = = = = = = = = =

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡ÖÂ
Àñ£¡
 
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡²ÌÝíÝí
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cbeibei <at> 263.net
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2004-12-03
 
_______________________________________________
tsvwg mailing list
tsvwg <at> ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg
Picon
Favicon

RE: Reviews of RSVP extensions (here, there, or ... ?)

Hello,

>> My impression hasn't been that there's been a shortage of time to
>> present or discuss these drafts in TSVWG, but a low interest 
>> level.  I'm
>> not sure that a new group would address that.
>> 
>> That's an impression; I could be way off.

I remember a "Hum-test" was run during the WG.
Although I can't remember the exat words, I believe the question was
something like "is the RSVP work discussed in the TSVWG useful?"  or "is
the RSVP work discussed in the TSVWG of interest to this audience?".
I thought the "Hum-level" was clearly quite high. So my impression is
that we had settled on that question and that there is agreement in
progressing this work.
Am I missing something?

I think James' question is rather about whether this should be done
under TSVWG or in some other group. Personally, I would suggest keeping
it under TSVWG until there is a specific reason to change (e.g. there is
a shortage of time to present/discuss all the items under TSVWG).

Cheers

Francois

>> 
>> -- 
>> Ted Faber
>> http://www.isi.edu/~faber           PGP: 
>> http://www.isi.edu/~faber/pubkeys.asc
>> Unexpected attachment on this mail? See 
>> http://www.isi.edu/~faber/FAQ.html#SIG 
>> 
Bose, Pratik | 7 Dec 2004 18:57
Favicon

RE: Reviews of RSVP extensions (here, there, or ... ?)

Actually I do remember the hum-test that Francois mentions, and I also
recall that a counter hum-test was taken for the same question below
(for those in favor of not supporting this work here) and there was
barely a hum for that - so I am not sure whether the interest being low
is really correct. I am sure the minutes can confirm this info....

> -----Original Message-----
> From: tsvwg-bounces <at> ietf.org [mailto:tsvwg-bounces <at> ietf.org] 
> On Behalf Of Francois Le Faucheur (flefauch)
> Sent: Tuesday, December 07, 2004 4:48 AM
> To: Ted Faber; Jukka MJ Manner
> Cc: James Polk (jmpolk); tsvwg
> Subject: RE: [Tsvwg] Reviews of RSVP extensions (here, there, 
> or ... ?)
> 
> Hello,
>  
> >> My impression hasn't been that there's been a shortage of time to 
> >> present or discuss these drafts in TSVWG, but a low 
> interest level.  
> >> I'm not sure that a new group would address that.
> >> 
> >> That's an impression; I could be way off.
> 
> I remember a "Hum-test" was run during the WG.
> Although I can't remember the exat words, I believe the 
> question was something like "is the RSVP work discussed in 
> the TSVWG useful?"  or "is the RSVP work discussed in the 
> TSVWG of interest to this audience?".
> I thought the "Hum-level" was clearly quite high. So my 
> impression is that we had settled on that question and that 
> there is agreement in progressing this work.
> Am I missing something?
> 
> I think James' question is rather about whether this should 
> be done under TSVWG or in some other group. Personally, I 
> would suggest keeping it under TSVWG until there is a 
> specific reason to change (e.g. there is a shortage of time 
> to present/discuss all the items under TSVWG).
> 
> Cheers
> 
> Francois
> 
> >> 
> >> --
> >> Ted Faber
> >> http://www.isi.edu/~faber           PGP: 
> >> http://www.isi.edu/~faber/pubkeys.asc
> >> Unexpected attachment on this mail? See 
> >> http://www.isi.edu/~faber/FAQ.html#SIG
> >> 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg <at> ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg
> 
Kwok Ho Chan | 7 Dec 2004 18:58

Re: Re: Re: Re: some problemsaboutdraft-baker-diffserv-ba sic -classes-0 4

Cai Bei-Bei:
We are not saying AF3x is "higher priority" than CS3.
We are just indicating that the 2 service classes need to be treated differently
in parts of the end-to-end path.
We are not saying how many queues and what the queue scheduling algorithm must be.
Hence we are not saying AF3x must be scheduled ahead of CS3.
-- Kwok Ho --

At 09:43 AM 12/7/2004, =?gb2312?B?sszd7d3t?= wrote:
Kwok Ho :
 
Thank your help!
 
In your previous letter, you said :"We have found that the Loss Tolerance for these traffic(Broadcast Video service)
 is relatively lower than Multimedia Streaming service class." Does this sentence mean that the broadcast video service need
lower packet loss than Multimedia Streaming service class? If it is yes, I think it is more reasonable that
broadcast video service have a higher priority. But broadcast video servie is CS3,whose priority is lower than
than Multimedia Streaming service(AF3).Do it have some antinomy?
 
Thanks a lot!
best regards!
                                                                  cai Bei-Bei
 
¡¡¡¡
 
======== 2004-12-06 23:28:11 ÄúÔÚÀ´ÐÅÖÐдµÀ£º ========
 
Bei-Bei:
Please notice I have added this thread to the TSVWG mailing list.
Please see response embedded below.
Any additional input are welcomed.
Thanks!
-- Kwok Ho --

At 10:39 PM 12/4/2004, =?gb2312?B?sszd7d3t?= wrote:
Kwok Ho:
 
Thank you very much for your so detailed answer.
Your help have resolved some questions I had.
Of course,you can put this conversation back onto the TSVWG mailing list.:)
 
And,may I understand the draft as follow : the gradation of traffic priority from high to low is
 CS7,CS6.EF,CS5.AF4,CS4,AF3,CS3,AF2,CS2,AF1,CS0(DF),CS1?

I think you can say that.  But in reality, IMHO, the network needs to make sure it
upholds the promise it gave to its customer, no matter which service class it is.
This may mean the network needs to admit less traffic that goes into service class using
EF so the service classes using CS4 and AF4x will still uphold the promise to its users.
The decision is on which service classes a network supports (they don't need to support
all the service classes) and how much of each.

I have cleared about those service classes except for the Broadcast Video Service Class.
Based on the draft,Broadcast video is inelastic traffic,and Multimedia Streaming is elastic.
Why do the former use CS3 and the latter use AF3?
And in 4.2.1 part of draft-baker-diffserv-basic-classes-02, the Multimedia Streaming class include
the Broadcast video service.What is the reason for separating the broadcast video from the multimedia streaming
in draft-baker-diffserv-basic-classes-03(and 04) ?

Broadcast Video service class is meant for High Definition broadcasts and the likes.
We have found that the Loss Tolerance for these traffic is relatively lower than Multimedia
Streaming service class.  Hence its separation.  Again this is based on what the application
requires of the network.  At low loss high capacity parts of the network, both of these
service classes may get treated together without impact to the promises made to the
user of each service class (discussion for the service class aggregation draft). 


Thank you very much!
 
 
 
                                                                  Cai bei-bei
ps:Best wishes to you!
¡¡¡¡
 
 
======== 2004-12-02 23:49:52 ÄúÔÚÀ´ÐÅÖÐдµÀ£º ========
 
Bei-Bei:
Agree that they all need low delay, low jitter.
Other factors that affects the service classes placement are loss tolerance, traffic elasticity,
and packet size (and its variation).

Telephony uses EF. the Multimedia Conferencing uses AF because it is elastic traffic.
There is also the Real-Time Interactive class using CS4 that handles inelastic video traffic.
Please reference Figure 1 of
  http://www.ietf.org/internet-drafts/draft-baker-diffserv-basic-classes-04.txt

EF PHB normally get deployed with some kind of upper limit on the traffic it treats, either
with some policer, rate limiter, and/or some other admission control methods.
With this in mind, the EF traffic should not starve the AF4x and CS4 traffic.  Each should
forward the traffic as configured (one may not want to have too much EF traffic that will
add too much delay to the CS4 and AF4x traffic making them un-acceptable).  Without
having the elastic traffic adversely affecting the inelastic traffic.  Or having large packets
adding too much jitter to small inelastic traffic.

This draft takes the points of view of how each application using the different service classes
will receive the network treatment it requires (end to end), when traffic in multiple service classes
need to share the network.  When these traffic traverse both low and high capacity networks/links.

There is another draft that discusses how in high capacity networks/links, multiple service classes
can be aggregated together.  I have published -00 version of this draft but I am in the process of
updating this draft to add more content and making changes based on the comments received.
I would advise to wait for the -01 of draft-chan-tsvwg-diffserv-class-aggr.

I hope this help to answer/resolve the questions/issues you have.

Also, can I put this conversation back onto the TSVWG mailing list?

Thank you!
-- Kwok Ho --


At 08:17 AM 12/2/2004, =?gb2312?B?sszd7d3t?= wrote:
Kwok Ho Chan:
 
Thank you for your answers.
Yes,I am a student of Nanjing University of Posts and Telecommunications of China.
Now we are doing a project about end-to-end IP QoS.
 
Every service has different QoS requirement.So it will be classed, and deployed with DSCP, then be treated differently based on the different class.
Voice and Vedio require real-time, very low delay, very low jitter,which must use EF PHB.So I am
puzzled that Multimedia Conferencing use AF PHB. Services have lower priority ,they may wait in a queue for a longer time, which will bring higher delay. The same question exist between multimedia  streaming and broadcast video and so on.
 
We want to use the draft to class services, and deal with these based on the DSCP. 
 
¡¡¡¡Thank you very much!!
 
======== 2004-11-24 22:54:53 ÄúÔÚÀ´ÐÅÖÐдµÀ£º ========
 
Bei-Bei:
Thank you for your comments.
I think you mis-understood the concepts.
The basic-classes draft separates the traffic into different classes
based on their traffic characteristics.
It does not necessarily mean Multimedia Conferencing is lower than Telephony,
they just needs to be separated.
We also recommend some treatment approaches for each Service Class based
on each of their traffic characteristics.
Please look at (read) the draft this way and let us know if think differently.
Thanks!
-- Kwok Ho Chan --
p.s. Are you based in China?  It looks like a Chinese name to me.
      How may you want to use this draft?


At 09:18 AM 11/24/2004, =?gb2312?B?sszd7d3t?= wrote:

Hi:

  I have some problems about draft-baker-diffserv-basic-classes-04.

  1.Why is the priority of Multimedia Conferencing(eg.H.323/V2 Video conferencing) lower than telephone? The former(video) sometimes need very low delay,eg.CBR video.

  2 Usually,the signalling precede the corresponding service. Why does the service class signalling have a lower priority than telephone? And i also don't know why the QAM is behind Low Latency  data,and the broadcast video is behind multimedia streaming?In my concepte,broadcast video is include in multimeida streaming and need less delay.

 
 
  Thank you very much!!
       

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Cai Bei-bei
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cbeibei <at> 263.net
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2004-11-24

 

= = = = = = = = = = = = = = = = = = = = = =

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡ÖÂ
Àñ£¡
 
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡²ÌÝíÝí
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cbeibei <at> 263.net
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2004-12-02
 

= = = = = = = = = = = = = = = = = = = = = =

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡ÖÂ
Àñ£¡
 
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡²ÌÝíÝí
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cbeibei <at> 263.net
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2004-12-03
 

= = = = = = = = = = = = = = = = = = = = = =

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡ÖÂ
Àñ£¡
 
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡²ÌÝíÝí
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡cbeibei <at> 263.net
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2004-12-07
 
_______________________________________________
tsvwg mailing list
tsvwg <at> ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg

Gmane