Hi Andy,

I‘m struggling to make the connection, since draft-ietf-bess-service-chaining 
is specifically about how to do service chaining without needing a new protocol 
like NSH, so SFF labels would never be used.

In draft-ietf-bess-service-chaining , if MPLS transport were used between 
Service Function Forwarders, a label would be allocated for a SF interface when 
a route pointing to it is installed (by the controller). This would be 
advertised to the controller and from there sent to the forwarders with egress 
interfaces for the previous SF in the chain. The label would be used at the 
bottom of the MPLS stack as is done normally.


-914 886 2534

From: "Andrew G. Malis" <agma...@gmail.com>
Date: Tuesday, December 4, 2018 at 5:10 PM
To: "bess@ietf.org" <bess@ietf.org>, 
Cc: "draft-ietf-mpls-sfc-encapsulat...@ietf.org" 
Subject: Comment on draft-ietf-bess-service-chaining-06
Resent-From: <alias-boun...@ietf.org>
Resent-To: <r...@cisco.com>, <wsmac...@juniper.net>, <dh...@cisco.com>, 
<brunorijs...@gmail.com>, <mnapier...@att.com>, <thomas.mo...@orange.com>
Resent-Date: Tuesday, December 4, 2018 at 5:10 PM

I just read the new revision of draft-ietf-bess-service-chaining. Although the 
draft doesn't use the RFC 8300 NSH, it could very easily take advantage of 
features provided by the NSH (such as metadata) by adding NSH over MPLS as 
defined in draft-ietf-mpls-sfc-encapsulation to the list of encapsulations 
listed in section 2.5. And this draft provides an excellent label distribution 
mechanism for NSH over MPLS. It would make a lot of sense to add a reference to 
draft-ietf-mpls-sfc-encapsulation in the list of encapsulations in section 2.5.


BESS mailing list

Reply via email to