On 2010-04-21 19:11, Philip Homburg wrote:
> In your letter dated Wed, 21 Apr 2010 16:33:03 +1200 you wrote:
>> By contrast, a source host
>> never has this problem, because it knows by construction when a flow
>> ends (chances are, flow == socket).
>
> I'm wondering about the 'never' part. What if an application uses sendto()
> to communicate over UDP with multiple peers, and some of those peers happen
> to be located on a single remote host.
>
> How is the kernel supposed to keep track of those flows? Or does the
> application have to maintain flow-ids?
The latter, I think, in such a complex case. RFC 3697 says:
"To enable applications and transport protocols to define what packets
constitute a flow, the source node MUST provide means for the
applications and transport protocols to specify the Flow Label values
to be used with their flows."
That was because we intentionally did not define exactly what a flow is;
the IPv6 WG didn't want to do so at that time.
Brian
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
