On Wednesday 29 July 2009 03:52:26 you wrote: > On Tue, Jul 28, 2009 at 4:55 PM, Matthew > Toseland<toad at amphibian.dyndns.org> wrote: > > AFAICS no. I'm not saying we should go with a static only host, I'm saying > > we shouldn't > > make dynamic that which doesn't need to be, on grounds of CPU cost (which > > determines > > responsiveness in practice). Modern systems (even apache 2 to a reasonable > > degree) > > can serve static content ridiculously fast, that's not true of dynamic > > content. > > Oh come on now, this argument is just as implausible. Are you > seriously claiming that there will be a consequential difference in > speed (in terms of user experience) between a statically served page, > and a page that is dynamically generated in response to a simple test > of a HTTP header? > > Research I've seen is that response time only has a measurable effect > if its over 100ms, and only a perceptible slowness if its over 700ms. > I can't believe that, unless we're running our web server on an Atari > 800XL, that there will be any significant difference in response time > just because the page is dynamically generated.
VMs can be very slow, in terms of time-slicing CPUs. And Google bills by CPU time. In both cases, for any sensible implementation, the CPU cost of sending a static web page which is frequently accessed is essentially nil, and the cost of running a script is considerably more than that. However, if it is implemented by a redirect in the apache config it is probably fast enough. When we have had slashdots, we have often had performance issues even with a dedicated server (which is vastly more powerful than any VM or shared hosting service). Although these have been due to configured limits in apache, or possibly to logging to a database. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 835 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090729/04769cce/attachment.pgp>
