[Nottingham] How ISP shenanigans hampers your browsing experience.

Michael Quaintance penfoldq at penfoldq.co.uk
Mon Jan 5 22:54:06 UTC 2015

Hmmm... not an entirely fair comparison or reasoning.

Yes, ISPs are doing incredible amounts of deep packet inspection and other
"legal but not exactly good sport" tricks, but the tests on this website
are not comparing like with like.

>From the text at the bottom of the page itself, the HTTP test is using
straight HTTP/1.1 without compression or domain sharding [1] and comparing
this to SPDY HTTPS (with compression and SPDY negates domain sharding).

SPDY is a great improvement for the modern Web, and if you're logged in to
Google, and using a modern build of Firefox or Chrome/Chromium, you've been
using SPDY for a long time now. I like SPDY as a protocol and it's adoption
is generally a good thing, although I would prefer wider HTTP/2 adoption if
only to stop Google being in control of too much.

I cannot be bothered right now to recreate this test with the correct
comparisons (HTTP vs HTTPS, SPDY HTTP vs SPDY HTTPS, HTTP with all modern
improvements turned on vs SPDY HTTPS) but maybe I'll get around to that one

Certainly, the overhead of HTTPS everywhere is not as much as some
detractors claim and there are many with a vested interest in slowing the
adoption of decent quality encryption, but just as they speak with forked
tongues to make their point more convincing, this website is also
misrepresenting the truth for its own agenda.


[1] Domain sharding is a way to get around the requirement in the HTTP
standards to limit concurrent connections. The concurrent connections are
per-host so you simply spin up many more hosts, often simply as name-based
virtual hosts e.g. on the same server as www.example.com you can have
www1.example.com, www2.example.com, www3.example.com and split your
requests across them giving you four times the concurrency and loading the
page faster - it's not 4 times faster but it's an improvement. I used 4
here because I got bored with typing but the idea scales much larger
although each new server improves the load time by a smaller amount.

SPDY avoids the original problem that domain sharding was invented for so
domain sharding with SPDY actually hurts page load time. But it's simple
enough to avoid domain sharding with SPDY since sharding was a hack to get
around an IETF standard in the first place.

On Mon, Jan 5, 2015 at 12:52 PM, Jason Irwin <jasonirwin73 at gmail.com> wrote:

> "Let's encrypt the web!"
> "No! Things will slow down! Think of the poor CPU trashing away with all
> that complicated crypto-math-magicks!"
> Well, yes. On the face of it HTTPS should be slower. Handshake, encrypt,
> decrypt etc.
> Got to be slower.
> Makes sense, there's more work being done.
> And it would be true IF your ISP wasn't doing various bits of deep packet
> inspection, injecting cookies or other dastardly deeds.
> http://www.httpvshttps.com/
> On a quick test I just didny, HTTPS was 88% faster (09s Vs 7.7s).
> Yeesh.
> H.
> _______________________________________________
> Nottingham mailing list
> Nottingham at mailman.lug.org.uk
> https://mailman.lug.org.uk/mailman/listinfo/nottingham
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.lug.org.uk/pipermail/nottingham/attachments/20150105/1e7f8822/attachment.html>

More information about the Nottingham mailing list