Re: PaceBatch and pipelining

[Dare Obasanjo <kpako@xxxxxxxxx>]

--- Tim Bray <Tim.Bray@xxxxxxx> wrote:
> > >
> > If I wanted to post 10,000 new entries, I'd use a
> shell script and  
> > curl,
> > or maybe libwww-perl if the content varies.  How
> can it simplify the
> > programming model if I don't have to do any
> programming on the
> > client for either solution, and don't have to do
> any programming on
> > the server for the pipelining solution?
> 
> Here we're at the core of the question.  I think
> that the difficulty  
> of packing the 10K into a <batch> element and
> sending it once is not  
> qualitatively different from the script-ware
> necessary to do all the  
> individual HTTP dispatches, and I think that the
> result-processing  
> (track which ones fail) would be a *LOT* easier if
> they all came back  
> in one batch-response.

Difficulty for whom? Our server devs would HATE it if
a client could tie up server resources by requesting
that we perform 10,000 operations in sequence within a
single request instead of taking advantage of the
server's ability to parellelize requests by making
multiple requests. 

Also reading the W3C Note at
http://www.w3.org/Protocols/HTTP/Performance/Pipeline
it seems that the benefits of batching over using
multiple requests isn't clear cut. It seemed to take
some tuning of both server & client and expertise with
TCP to get the positive results from using pipelining
over using multiple requests.



THINGS TO DO IF I BECOME AN EVIL OVERLORD #222
I reserve the right to execute any henchmen who appear to be a little too intelligent, powerful, or devious. However if I do so, I will not at some subsequent point shout "Why am I surrounded by these incompetent fools?!"


		
____________________________________________________ 
Yahoo! Sports 
Rekindle the Rivalries. Sign up for Fantasy Football 
http://football.fantasysports.yahoo.com