Re: [tsvwg] A review of draft-ietf-tsvwg-udp-options-12

[Joseph Touch <[e-mail]>]

> On Jun 15, 2021, at 9:58 AM, Tom Herbert <[e-mail]> wrote:
> 
>> And, FWIW, trailers are just as easy if not *EASIER* for zero-copy to handle; they just DMA the user data first then the UDP options.
> 
> Joe,
> 
> I am dubious of that statement.

Zero-copy matters for the body, not the options.

For TCP, you have to start your DMA after the variable-length header.

For UDP with FRAG, the same thing happens. 
For UDP without FRAG, you start the DMA 8 bytes after the IP header.

> You'll have to show running code and
> performance results.

If the above won’t convince you, a demo won’t help.

> Software stacks and hardware have long been
> designed around the fact that networking protocols are in headers; UDP
> options is an outlier in that regard.

Again, not for FRAG; FRAG works exactly the same way.

And again, doing otherwise for non-FRAG is not a viable path forward due to legacy constraints.

> While it is certainly possible
> to make a functional stack to handle trailers, it is going to be hard
> if not impossible to leverage many implementation optimizations and
> techniques we've done over the years.

*That* statement really needs proof by running code and performance, IMO.

Either you can do DMA from an known offset for a known length or you can’t. 

Joe