Re: PaceBatch and pipelining

[Bill de hÓra <bill@xxxxxxxxxx>]

Mark Baker wrote:
> On Tue, Jun 28, 2005 at 08:36:48AM +0200, fmantek wrote:
> 
>>I like to bring up another argument in favor of a batch mechanism
>>build into the protocol.
> 
> 
> Just to put this in context...  by virtue of APP being HTTP/1.1 based,
> it already has a batching mechanism; HTTP pipelining.  PaceBatch would
> be a second batching mechanism.

They're mechanisms that seem to be optimizing different things.
Pipelining is an optimization to hold open a TCP connection as best I
can tell, PaceBatch is an optimization to hold open an APP aware process
as best I can tell. So, one's targeting the server i/o, anothers'
targeting the parser i/o ;) Depending on your server setup you can
expect entirely different benefits.

It would help if those arguing pro or con could tell us what they want
to optimize for, if anything - for all I know this is /purely/ a
convenience mechanism for the client to stuff entries together a la
Frank's point and has nothing much to do with the server-side handling
or connection details.

cheers
Bill