Re: Should Web Services be served by a different HTTP n+1?

[Phillip Hallam-Baker <[e-mail]>]

On Thu, Jan 24, 2013 at 7:44 PM, Robert Brewer <[e-mail]> wrote:

> Roberto Peon wrote:
> > The worst part is the high latency, especially given TCP's
> > current cogestion avoidance implementations-- the total number
> > of round-trips ends up dominating latency, regardless of how
> > much bandwidth one has.
>
> Why is this being addressed by trying to make the messages smaller?
> Wouldn't following the original architecture of HTTP, which was optimized
> for fewer, larger messages, also reduce latency?
>

Well this all goes back to Tim's original point that maybe we should have
had more than 6 hours discussion before a grad student just threw in the
<IMG> tag because he thought it would be cool and Eric had already written
the code for images in the widget.


When we moved from one request per TCP session to multiple requests per TCP
session it was not exactly done well. Each request is still formatted as if
it was running in a separate session and so it has to reconstruct the whole
browser context.

I don't see why we could not do a delta encoding on the headers as follows

Old:

Get /foo
Header1:
Header2:
Header3:

Get /bar
Header1:
Header2:
Header3:


New:

Get /foo
Fred:[
Header1:
Header2:
Header3:
]

Get /bar
#include Fred


The client knows when headers are likely to be repeated across requests in
the same session - they are mostly repeats anyway.



-- 
Website: http://hallambaker.com/