Date: 26 Dec 2001 11:45:58 -0500
From: "Perry E. Metzger" <[EMAIL PROTECTED]>
Message-ID: <[EMAIL PROTECTED]>
| I'll repeat (broken-record like), we've already deployed the end point
| implementations in the field. We can't change those stacks at this
| point.
There are two points there. First, as Keith said, this is nonsense,
nothing that is being discussed here has any impact upon deployed nodes
(other things that could be discussed would, this doesn't). The very worst
is that users of currently shipping OS's might not be able to take advantage
of some benefit in the future, and might need to upgrade. But that's hardly
a new concept.
But if what you say were true, the result would be that you'd be arguing
for exactly the opposite of what you seem to believe that you are. That is,
assuming that XP has implemented an interface anything like that defined in
2553 (and while I've never seen an XP, I'd be surprised if they hadn't)
then the ability to set the flow label already exists.
That is ...
struct sockaddr_in6 {
sa_family_t sin6_family; /* AF_INET6 */
in_port_t sin6_port; /* transport layer port # */
uint32_t sin6_flowinfo; /* IPv6 traffic class & flow info */
struct in6_addr sin6_addr; /* IPv6 address */
uint32_t sin6_scope_id; /* set of interfaces for a scope */
};
That means, that if we were to make the field "reserved, must be zero"
then there's no way the shipped implementations could comply, as they will
probably already allow applications to set the value to anything they want.
On the other hand, if we (or someone) simply define a meaning for the bits,
then all that should be required is for someone to install an application
that uses them in the defined way.
That is, the "already shipping" argument in this case has the exact
opposite effect than you're claiming that it does.
Now for the already shipping IPv6 routers things are different - but the
volume of those is on a different plane, and they tend to be operated by
people who also know how to upgrade them to newer code if that turns out
to be needed to support something they want to offer. That is, if using the
flow label is needed (or useful) to support some QoS paradigm, and the
router (or its owner) wants to support that, then making the router do it
won't be at all impossible.
kre
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
