Re: [Openv6] Starting discussion

["Guang Yao" <[e-mail]>]

Hi Will and folks,

 

I want to introduce SAVI and discuss whether it can be adopted in the OpenV6
framework. 

 

SAVI(Source Address Validation Improvements, http://tools.ietf.org/wg/savi/)
sets up bindings between IP address and binding anchor. It has following
features:

 

1.       There are at least 3 SAVI solutions for IPv6: SAVI-DHCP,SAVI-FCFS,
SAVI-SEND. The solutions are designed based on the address assignment
mechanism (DHCPv6, Stateless, SEND-CGA). If new address assignment
mechanisms (or address usage guideline) are introduced, e.g., IKEv2, Mobile
IPv6, Ipv4/IPv6 transition, corresponding SAVI solutions should be
implemented.

 

2.       In general, only one of the SAVI solutions should be enabled,
depending on the deployment scenario.

 

3.       Multiple SAVI solutions can coexist on the same device. For
example, SAVI-DHCP and SAVI-FCFS may work together for the scenarios where
both DHCP and stateless addresses are allowed.

 

4.       SAVI solutions have the same data plane model (or abstraction):
binding table. Binding table contains the valid associations between IP
address and the corresponding binding anchor.

 

5.       The differences of SAVI solutions are in the control plane.
Difference SAVI solutions snoop difference protocols, and even data packets.

 

6.       SAVI solutions can benefit from a global view of the network,
because a protection perimeter composed by multiple SAVI devices can be
automatically formed.

 

We have implemented a prototype of Openflow-based SAVI. It is quite simple
and effective, but it's a pity that it cannot work with non-Openflow
devices. Besides, currently it is hard for the Openflow-based SAVI to
cooperate with other applications, e.g., transition applications. We are
interested in whether an Openv6-based SAVI can have more advantages than the
Openflow-based one? (it seems more general interfaces are supported?)

 

Best regards,

Guang

 

 

From: Liushucheng (Will) [mailto:[e-mail]] 
Sent: Monday, January 27, 2014 4:16 PM
To: [e-mail]
Subject: Starting discussion

 

(Sorry for duplicated mails.)

 

Hello,

 

Thanks for joining the Open IPv6 (openv6) mailing list. It's great to see so
much interest in the topic!

 

The following questions are listed to be discussed here: 
(1) What are the problems and use cases existing in various IPv6
applications, e.g., multiple IPv6 transition technologies co-exist? 
(2) How to enable the applications to program the equipment to tunnel IPv6
traffic across an IPv4 data plane? 
(3) How this work can be done through a general interface, e.g., to
incorporate the transition policies, simplifying the different stages
through the transition and guaranteeing that current decisions do not imply
a complicated legacy in the future? 
(4) How to make the end-to-end configuration of devices: concentrator/CGN,
CPE and the provisioning system? 
(5) How to extend the existing IETF protocols, e.g., netconf, to support
this open interface?

 

In addition to the above questions, we've laid out an initial description of
the problem in
https://datatracker.ietf.org/doc/draft-sun-openv6-problem-statement/ and are
looking for productive discussion and feedback on this list. We are now
modifying it and a new version will be released accordingly. :)

 

Regards,

Shucheng LIU (Will)

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

Shucheng LIU (Will), Ph.D.

Researcher & Standard Representative 

Huawei Technologies Co.,Ltd

[e-mail] <mailto:[e-mail]> 

http://www.linkedin.com/in/shucheng

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