Re: [v6ops] WG Doc? draft-gont-v6ops-ipv6-ehs-packet-drops

[Tom Herbert <[e-mail]>]

On Mon, Mar 14, 2016 at 10:59 AM, Nick Hilliard <[e-mail]> wrote:
> Tom Herbert wrote:
>> From that point of view, the IPv6 flow label "works best" for ECMP
>> and so the three tuple hash (addresses and flow label) should be the
>> recommended method for ECMP devices to implement.
>
> Tom,
>
> Everyone agrees with this.  The problem is that core and leaf operators
> are not in control of flow label generation at the edge, and many
> operating systems do not bother to set the flow label.  In other words,
> flow labels do not currently provide enough entropy to be useful for
> flow hash generation in most instances.
>
> Everyone understands that flow labels would help this problem.  This
> draft is about what happens when the flow label is zero, which is the
> case more often than not.
>

Nick,

iOS already sends non-zero IPv6 flow labels, we have changed Linux
default to use IPv6 flow label so once Android rebases the kernel the
majority of IPv6 end hosts should be sending non-zero flow labels. It
is far easier and shorter time to change the software stacks to send
them than upgrading all the HW in the paths to use them productively.
But compared the problem of HW being able to efficiently parse
arbitrarily long header chains, I still think this is a very solvable
problem.

Tom

> Nick