RE: Reducing memory footprint?

• Messages sorted by: [ date ][ thread ][ subject ][ author ]
• Next message: Paul Salaberria: "parsing tables"
• Previous message: Nick Ing-Simmons: "Re: Reducing memory footprint?"

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_000_01BE6AF2.39061F48
Content-Type: text/plain;
	charset="iso-8859-1"

Just a thought: do your 3000 users really need 3000 simultaneous processes
in order to simulate the load ? If they are not all likely to be active
simultaneously, then perhaps you could consider running fewer  processes,
but reducing the wait times to increase the effective load.

It should be possible to demonstrate that this can produce the same loading,
by comparing the two scenarios for various smaller volumes.

Not sure if this could be relevant, but another approach which we have used
in a very simple application (send Telnet message; collect response later)
is to write a mini round-robin scheduler that uses a single process. [Open
all connections; send message on each; wait for first reply; process reply
and send another message. Repeat from wait.] This depends on being able to
check which connections have data ready, and is probably rather hard work
for a more complicated scenario such as yours where each connection has
several stages to process. [We used IO:socket, and the select() call for
checking for data readiness.]

HTH,
Sebastian Bazley

------_=_NextPart_000_01BE6AF2.39061F48
Content-Type: application/ms-tnef
Content-Transfer-Encoding: base64
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------_=_NextPart_000_01BE6AF2.39061F48--

• Next message: Paul Salaberria: "parsing tables"
• Previous message: Nick Ing-Simmons: "Re: Reducing memory footprint?"