Re: Pipelining and compression effect on HTTP/1.1 proxies

On Tue, 22 Apr 1997, Benjamin Franz wrote:

> majority of people browsing are doing so over modem links that *already*
> perform pretty good on the fly compression of the data flowing through
> them - thus reducing the potential savings to the end user from
> pre-compressing text/* to negligible.

But the user's modem link is only one of many hops between the user and
the server so I don't agree that the fact of modem compression on one hop
negates the importance of compression of the original material. 

Another point to consider is that typical content travels over several
hops before arriving at the client. Depending on content length, 
hop-link speeds, hop-queues, etc. a 7.5% reduction in effective length
can have a much greater effect on the transit time from the server to
the client.

And as has already been pointed out, the content most impacted by end-end
compression is likely to be the HTML controlling the remainder of the
presentation. My sense it that the next transition in web traffic is
likely to be to much larger and more complex 'html' files as providers add
complex J(ava)Script et al to the pages. This would counter balance the
most recent shift we've seen toward more graphical content.

>From my perspective, there is a very real payback here for very minimal
implementation effort.

Dave Morris

Received on Tuesday, 22 April 1997 14:52:43 UTC