Re: Reference set in HPACK

[Roberto Peon <[e-mail]>]

I'm all for people generating such data and coming to conclusions based on
it.
I'd caution that one should not predicate the analysis on pages that look
exactly as they do today, nor headers which look exactly as they do today.

Essentially all of the time that I commit to HTTP2 these days comes out of
personal time, and doing such an analysis would require more time than I'm
willing to take away from my family or more likely, sleep.

-=R


On Wed, Jul 2, 2014 at 12:37 AM, Greg Wilkins <[e-mail]> wrote:

>
> On 2 July 2014 08:09, Roberto Peon <[e-mail]> wrote:
>
>> Even a few bytes per request adds up as the number of requests go up.
>> Lets think about a page today that has 100 elements, which is something
>> we see today.
>> If the headers were regularized, with the use of a reference set, one
>> could imagine a reduction of 20 bytes per header.
>> With 100 elements, this is approximately 2k of data, or two packets worth.
>>
>
> Roberto,
>
> I think this kind of calculation needs some real data.     Maintaining a
> reference set is not a zero byte deal, as it still requires bytes to be
> sent to remove entries from the reference set that are not needed.  If real
> data shows that removals are frequent, then they may add up to close to the
> 2k also.
>
> I don't think that it is unreasonable to request that realistic data is
> provided to prove that a reference set is worthwhile.   It  would be good
> to obtain and evaluate such data for both a single connection and coalesced
> connections.
>
> regards
>
>
>
>
>
>
>
>
> --
> Greg Wilkins <[e-mail]>
> http://eclipse.org/jetty HTTP, SPDY, Websocket server and client that
> scales
> http://www.webtide.com  advice and support for jetty and cometd.
>