RE: AW: [ANCP] Layer 2 Parameters in the Protocol document

[Michel.Platnic at ecitele.com]

Hi Sanjay,

In case of multiple VC or VLAN for which
we need to send the encapsulation as part of the 
'discovered parameters', I agree with
you that best method should be to allow 
one message to transport several Access-Loop-Remote-IDs.
I guess we should probably add one sentence
in the draft to document this case.

By the way, the 'bulk transaction message'
is defined in draft-ietf-gsmp-v3-base-spec-06
but not in RFC3292, do we want to have
references to a draft that is expired?

About the other point: wether or not
to have the vpi.vci parameters as optional
within the Access-Loop-Remote-ID, if
we do decide to keep these parameters
for historical reasons, we should probably
define a default value to be used
for dummy vpi.vci.

Thanks and Regards,
Michel.

"Sanjay Wadhwa" <swadhwa at juniper.net>

27/09/2006 21:42

To	<Michel.Platnic at ecitele.com>, "Haag, T" <Thomas.Haag at t-systems.com>
cc	<ancp at ietf.org>
Subject	RE: AW: [ANCP] Layer 2 Parameters in the Protocol document

Hi Michel
  One of the main requirements on the
BNG is to be able to correlate L2C messages pertaining to a given access
loop with subscriber sessions over that access-loop. This is done via matching
ACIs received in L2C messages with what is received in DHCP opt 82 or PPPoE
IA tag.  Ideally the BNG should not have to interpret these values
and should be able to to transparently do an exact match. Keeping the formats
in L2C and DHCP/PPPoE identical allows this. 
The IETF draft mentions the syntax that you
refer below as "typical" syntax. However, it clearly states that
the format of the ACI should be identical to what is inserted by the DSLAM
(signalling relay) in subscriber signalling messages (as defined in TR-101,
cited as a reference in the IETF draft). TR-101 specifies a default syntax
for the ACI which includes VPI/VCI. However, it allows the default syntax
to be overriden by "flexible" syntax, which I believe lets you
specify just the slot/port.
The intent in the IETF draft it to only require
that the ACI sent in L2C and DHCP/PPP messages by the DSLAM should match
(and the format should be transparent to the BNG).
 
To your other point about encaps for multiple
VCs per access-loop, it doesn't have to be a seperate message. It could
be a single message containing a set of {ACI, DSL line attributes}. Other
then encaps, rest of the attributes would be identical in each set.
 
Regards
-Sanjay
 
 
  
-----Original Message-----
From: Michel.Platnic at ecitele.com [mailto:Michel.Platnic at ecitele.com]
Sent: Wednesday, September 27, 2006 12:18 PM
To: Haag, T
Cc: ancp at ietf.org
Subject: Re: AW: [ANCP] Layer 2 Parameters in the Protocol document


Hi Thomas and All, 

Layer 2 parameters transmitted within the Access-Loop-Circuit-ID are indeed
a MUST for several use cases, 
so, as explained in your email, this layer 2 information should not be
removed. 

The Access-Loop-Circuit-ID is defined within 'draft-wadhwa-gsmp-l2control-configuration-01.txt'
as follows: 
 “Access-Node-Identifier atm slot/port:vpi.vci”   (when ATM/DSL
is used)

 “Access-Node-Identifier eth slot/port[:vlan-id]” (when Ethernet/DSL is used)

This means that for ATM interfaces at UI, vpi.vci MUST always be transmitted.


This brings us to at least 3 scenarios for the topology discovery use case (when ATM/DSL is used):
1- no layer 2 information needed, only layer 1 information required to be transmitted
2- single VC and layer 2 encapsulation required to be transmitted
3- multi-VC and layer 2 encapsulation required to be transmitted

Comments:
Scenario1: vci.vpi is actually not required to be transmitted within Access-Loop-Circuit-ID
Scenario2: vci.vpi needed
Scenario3: if encapsulation is needed for each (n) vci.vpi, it means that n topology discovery messages must be transmitted.



Proposal:
If we replace  “Access-Node-Identifier atm slot/port:vpi.vci”   (when ATM/DSL is used)
by  “Access-Node-Identifier atm slot/port:[vpi.vci]”   (when ATM/DSL is used)
then vpi.vci becomes optional and we are fine with case 1.

Question:
Are we fine with the fact that in case of multiple VC (multiple VLAN as well) we must transmit multiple messages?


Regards,
Michel.

"Haag, T" <Thomas.Haag at t-systems.com>

26/09/2006 23:38

To	<Menachem.Dodge at ecitele.com>
cc	ancp at ietf.org
Subject	AW: [ANCP] Layer 2 Parameters in the Protocol document

Menachem, 
  
The usage of a VPI/VCI is really dependent on the use case you are considering.
Regarding the use case access line topology it is optional because there
is no strong relation to a layer 2 parameter if we report the net data
rate. 
  
But considering the use cases OAM (F5 LB for ADSL2/2plus) or line configuration
the layer 2 parameter such as VPI/VCI or VLAN is essential. So I would
recommend keeping these layer 2 parameters in the format of the “Access-Loop-Circuit-ID”.
  
Regards 
  
Thomas

---

Von: Menachem.Dodge at ecitele.com [mailto:Menachem.Dodge at ecitele.com]
Gesendet: Montag, 28. August 2006 15:57
An: ancp at ietf.org; swadhwa at juniper.net; jmoisand at juniper.net; ssubramanian at juniper.net;
Haag, Thomas; norbert.voigt at siemens.com
Betreff: [ANCP] Layer 2 Parameters in the Protocol document
  

Hi, 

       I have a question regarding the "Access-Loop-Circuit-ID"
as defined in the document "draft-wadhwa-gsmp-l2control-configuration-01.txt".

       In the case of an Ethernet based local loop the format
is defined as: 

       "Access-Node-Identifier eth slot/port[:vlan-id]"
meaning that the vlan-id is optional.

       In the case of an ATM based local loop the format
is defined as: 

       "Access-Node-Identifier atm slot/port:vpi.vci"
 meaning that the vpi.vci is mandatory.

       I would like this changed to "Access-Node-Identifier
atm slot/port[:vpi.vci]" meaning that the vpi.vci is optional.

       This would allow for multiple vpi/vci(s) on the same
port as a message could be sent  without the vpi.vci field.


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

I have a similar question regarding the "Access-Aggregation-Circuit-ID-ASCII"
as defined in the document "draft-wadhwa-gsmp-l2control-configuration-01.txt".

       In the case of an Ethernet based local loop the format
is defined as: 

       "Access-Node-Identifier eth slot/port[:inner-vlan-id][:outer-vlan-id]"
meaning that the vlan-ids are optional.

       In the case of an ATM based local loop the format
is defined as: 

       "Access-Node-Identifier atm slot/port:vpi.vci"
 meaning that the vpi.vci is mandatory.

       I would like this changed to "Access-Node-Identifier
atm slot/port[:vpi.vci]" meaning that the vpi.vci is optional.

       
Thank you kindly. 

Best Regards, 
Menachem _______________________________________________
ANCP mailing list
ANCP at ietf.org
https://www1.ietf.org/mailman/listinfo/ancp

_______________________________________________
ANCP mailing list
ANCP at ietf.org
https://www1.ietf.org/mailman/listinfo/ancp