RE: RFC3697 question,can anybody explain it ?

[Lawrence Zou <zou.rong@huawei.com>]

>As the RFC says, it doesn't describe any mechanisms for establishing
flow state, and therefore it doesn't need to describe the data either.
But I would assume that the flow state will >always include at least
>  Flow label value
> Source address
>   Destination address
>with the rest depending on the way the flow label is being used.


Brian:
    thanks, it helps me a lot,but i assume that except that 3- tuple you
mentioned,the flow state will at least also include the  liftetime of
the flow label. 
  

-----Original Message-----
From: Brian E Carpenter [mailto:brc@zurich.ibm.com] 
Sent: Tuesday, May 16, 2006 10:43 PM
To: Lawrence Zou
Cc: ipv6@ietf.org; v6ops@ops.ietf.org
Subject: Re: RFC3697 question,can anybody explain it ?


Lawrence,

Lawrence Zou wrote:
> what the definition of the "flow state" is? I am not very clear about 
> the content of "flow state" and there is no definition  in the 
> RFC3697.

I assume you have read the definition of "flow" at the beginning of
section 1.

"State" or "state information" in the IETF usually means the set of data
held by a node to describe a communications session. So in this case, a
possible definition of flow state would be:

   The set of data held by a node to describe a flow as defined in
section 1.

I'm sorry we didn't include a definition in the RFC.

As the RFC says, it doesn't describe any mechanisms for establishing
flow state, and therefore it doesn't need to describe the data either.
But I would assume that the flow state will always include at least
   Flow label value
   Source address
   Destination address
with the rest depending on the way the flow label is being used.

Hope this helps.

     Brian