The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2000-May> msg00393



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Re: MPLS - ATM interworking?

  • From: Shahram Davari <Shahram_Davari@pmc-sierra.com>
  • Date: Wed, 31 May 2000 14:03:36 -0700
  • Cc: mpls@UU.NET, "'Andrzej Czerczak'" <Andrzej_Czerczak/HeadQ@netia.pl>, rraszuk@cisco.com

Hi,
 
As I said there is currently no ECN application in MPLS networks, rather
there is ECN processing in MPLS networks for TCP applications. 
 
I think you agree that ECN is useful because it can control congestion
without dropping packets, which leads to higher BW utilization (i.e., does
not waste BW with the packets which are dropped). This goal is achieved  by
reducing the TCP transmission window. 
 
If I understand correctly, what you are proposing is that an ingress LSR may
be able to use the ECN notification to divert some of the traffic to other
LSPs (i.e., balance the load across multiple LSPs), which leads also to
higher BW utilization. In other words ECN notification may be as an input to
the Constraint-Based routing algorithm, which runs on the ingress LSRs. 
 
However, the same objective can be achieved by feeding back the current
congestion status information in LSRs to the ingress LSRs, using the LSP
feedback mechanism (check out section 7 of
http://www.ietf.org/internet-drafts/draft-ietf-mpls-te-feed-00.txt).
<http://www.ietf.org/internet-drafts/draft-ietf-mpls-te-feed-00.txt)>  , or
even with extensions to IGPs. Using this mechanism you won't need the ECN
bits and ECN notification for the edge-to-edge congestion control, because
the congestion information is available locally on each LSR and the
feed-back is also generated locally.
 
Does this answer your question?
 
Regards,
-Shahram
 
 
 
 
 
 

[Shahram Davari] 
 -----Original Message-----
From: CATANZARITI Sergio FTR&D/TI
[mailto:sergio.catanzariti@rd.francetelecom.com]
Sent: Wednesday, May 31, 2000 4:01 PM
To: 'Shahram Davari'
Cc: mpls@UU.NET; 'Andrzej Czerczak'; rraszuk@cisco.com
Subject: RE: Re: MPLS - ATM interworking?



Thanks, I knew that RFC. But, it does not solve the original question that
was ECN application in an MPLS network and its usage.

Sergio 

	--------------------------------------------------------------------

Sergio Catanzariti 
Senior Project Manager, Technology Integration 
France Telecom R&D 
1000 Marina Boulevard Suite 300 
Brisbane CA 94005 
Tel. 650-875-1526 
Fax. 650-875-1505 
email:sergio.catanzariti@rd.francetelecom.com 
-------------------------------------------------------------------- 


	-----Original Message----- 
From:   Shahram Davari [SMTP:Shahram_Davari@pmc-sierra.com] 
Sent:   Wednesday, May 31, 2000 12:50 PM 
To:     'CATANZARITI Sergio FTR&D/TI'; Shahram Davari 
Cc:     mpls@UU.NET; 'Andrzej Czerczak'; rraszuk@cisco.com 
Subject:        RE: Re: MPLS - ATM interworking? 

	Hi, 
  
1) Please check RFC 2481 for the definition and usage of ECN. 
2) As far as I know there is currently no proposal to use ECN bits in an 
Edge-to-Edge level. The intention of Sally Floyd's draft was that ECN to be 
used with TCP (end-to-end). 
  
Regards, 
Shahram 

	-----Original Message----- 
From: CATANZARITI Sergio FTR&D/TI 
[ mailto:sergio.catanzariti@rd.francetelecom.com
<mailto:sergio.catanzariti@rd.francetelecom.com> ] 
Sent: Wednesday, May 31, 2000 3:13 PM 
To: 'Shahram Davari' 
Cc: mpls@UU.NET; 'Andrzej Czerczak'; rraszuk@cisco.com 
Subject: RE: Re: MPLS - ATM interworking? 



	Shahram, 

	        the draft-ietf-mpls-ecn-00.txt does not say anything about
ECN usage 
and applications, but only about ECN support and definition in an MPLS 
network, the things are quite different. So, I have the same question: where

I can find Ramakrishnan et al., "A Proposal to Incorporate ECN in MPLS", 
draft-ietf-mpls-ecn-00.txt? 

	        Also, I do believe that the ECN in an MPLS network works on
resource 
types (LSPs) quite different than native IP flows. So, the source address is

not that useful here, albeit in an ECN schema in pure IP networks. 

	        ECN in an MPLS network is supposed to signal congestion in
an 
edge-to-edge fashion (different from end-to-end) where the LERs could 
cooperate on that. 

	        Absolutely, I want to associate LSPs with congestion
dynamics to 
eventually slow them down. How can I do that? That's the question. 

	Sergio 

	
-------------------------------------------------------------------- 

	Sergio Catanzariti 
Senior Project Manager, Technology Integration 
France Telecom R&D 
1000 Marina Boulevard Suite 300 
Brisbane CA 94005 
Tel. 650-875-1526 
Fax. 650-875-1505 
email:sergio.catanzariti@rd.francetelecom.com 
-------------------------------------------------------------------- 


	        -----Original Message----- 
From:   Shahram Davari [SMTP:Shahram_Davari@pmc-sierra.com] 
Sent:   Wednesday, May 31, 2000 11:34 AM 
To:     'Andrzej Czerczak'; rraszuk@cisco.com 
Cc:     mpls@UU.NET 
Subject:        RE: Re: MPLS - ATM interworking? 

	        Andrzej, 

	        1)ECN usage is defined in draft-ietf-mpls-ecn-00.txt, which
is 
attached to 
this email. 

	        2)Congestion Notification is and end-to-end signal which is
meant to 
be used 
by TCP. You don't need to know the ingress LSR or even the LSP from which 
the congested packet is received. The receiver may send the congestion 
notification to the sender using source IP address, either through another 
LSP or even through normal IP hop-by-hop routing. 

	        Regards, 
-Shahram 




	        >-----Original Message----- 
>From: Andrzej Czerczak [ mailto:Andrzej_Czerczak/HeadQ@netia.pl
<mailto:Andrzej_Czerczak/HeadQ@netia.pl>  
< mailto:Andrzej_Czerczak/HeadQ@netia.pl
<mailto:Andrzej_Czerczak/HeadQ@netia.pl> > ] 
>Sent: Wednesday, May 31, 2000 1:46 PM 
>To: rraszuk@cisco.com 
>Cc: mpls@UU.NET 
>Subject: Re:Re: MPLS - ATM interworking? 
> 
> 
> 
>Robert, 
>First of all thanks for your comments. 
> 
>>Have you read draft MPLS Support of Differentiated Services 
>>draft-ietf-mpls-diff-ext-04.txt where we are proposing to use the ECN 
>>(explicit congestion notification) on a per LSP basis using 
>the EXP bits 
>>in the shim header ? Currently this is the most attractive 
>option, but I 
>> am not aware of any implementation support this today. 
> 
> 
> 
>a) Document draft-ietf-mpls-diff-ext-04.txt does not not 
>define the usage 
>of ECN. 
> 
>b) Taking into account label merging capabilites of LSRs, how 
>can determine 
>to which ingress LSP to to send Congestion Notification? 
> 
>c) Where can I find: 
> 
>Ramakrishnan et al., "A Proposal to Incorporate ECN in 
>   MPLS", draft-ietf-mpls-ecn-00.txt? 

	        > 
> 
> 
>What I'd like to say is, that at least during first phases of 
>implementations (before entire infrastructure will become IP) 
>interworking 
>of MPLS core and ATM edge will be important, but I could not find a 
>proposal of how to do interworking between them. 
> 
> 
> 
>Andrzej 
> 
> 
> 
> 
> 
> 
 << File: draft-mpls-ecn-00.txt >>