Re: Getting to consensus on packet number encryption

[Ian Swett <[e-mail]>]

>From my perspective, it's just "more work" and certain workloads are really
CPU cost sensitive in datacenters in a way that isn't true on the public
internet.

That being said, datacenters may start moving to large MTUs earlier than
the public internet, which would greatly decrease the practical cost of PNE.

Given where we're at, even though I believe it's likely someone will want
to disable PNE for some use case at some point, I think we should spec PNE
as the only option for now and when someone needs to disable it, it can be
up to them to create a mechanism, use it in their own implementation, and
if they care to standardize it, they can?

On Thu, Apr 26, 2018 at 8:38 AM Patrick McManus <[e-mail]>
wrote:

>
>
> On Wed, Apr 25, 2018 at 8:21 PM, Ian Swett <
> ianswett=[e-mail]> wrote:
>
>> It has been, and I'm personally supportive of it, because I believe it'll
>> be useful for datacenter QUIC use cases.
>>
>
> can you layout the datacenter and PNE mismatch? It is just "more work" or
> is it an interaction with other hardware offloads?
>
>
>