W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2014

Re: Will HTTP/2.0 be green ?

From: Mike Belshe <mike@belshe.com>
Date: Sun, 1 Jun 2014 15:40:21 -0700
Message-ID: <CABaLYCvXEvNMO4s0q8Bwh8Y2-OB3tnqYKZ5DGf=-4v+=JZqPDQ@mail.gmail.com>
To: Greg Wilkins <gregw@intalio.com>
Cc: Poul-Henning Kamp <phk@phk.freebsd.dk>, HTTP Working Group <ietf-http-wg@w3.org>
On Sun, Jun 1, 2014 at 3:12 PM, Greg Wilkins <gregw@intalio.com> wrote:

>
> On 1 June 2014 23:32, Mike Belshe <mike@belshe.com> wrote:
>
>> I would be interested in seeing a useful study here.  However, the best
>> study would be one done in production environments rather than on simulated
>> devices with simulated models.  It would just be really hard to get a
>> realistic model.
>
>
>
> Mike,
>
> It would be interesting to know if any large scale adopters of SPDY have
> seen any significant changes in power usage in production environments - at
> least on the server side.
>
> There are significant increases in CPU demands with SPDY, but I would also
> guess a lot of benefits as well.
>

I don't know if there are net power increases with HTTP/2 or not.

Some thoughts:

a) In the datacenter, we can generally stipulate that more CPU correlates
to more power draw.  However, in these environments, the web servers are
such a minuscule portion of the aggregate power draw that this is unlikely
to be a significant difference.

b) On handheld devices, we know that the largest draw of power is generally
the radio.  The reduced number of connections and packets in HTTP/2 will be
a significant win for the radio such that overall power consumption may be
significantly less than that with plain HTTP, in spite of using more CPU.
 Implementation optimizations (number of streams, compression parameters,
timing, etc) likely have more effect on power usage than HTTP1 vs HTTP2.
 Either way, I don't know how you'd isolate implementation-specific power
draw vs "protocol" power draw.



>
> While I tend to agree that this is not something we will be able to very
> accurately measure nor to sweat any small changes, it would probably be
> worthwhile for somebody with the right equipment to just verify we are not
> making any huge increases in power requirements.
>


A measurement would be great.   But I don't think we'll have a good
analysis of this before HTTP2 in this form (or any other form) completes.


Mike






>
> cheers
>
>
>
>
>
> --
> Greg Wilkins <gregw@intalio.com>
> http://eclipse.org/jetty HTTP, SPDY, Websocket server and client that
> scales
> http://www.webtide.com  advice and support for jetty and cometd.
>
Received on Sunday, 1 June 2014 22:40:49 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:14:31 UTC