Mike wrote: > Erik, > > thaks for the reply. I conducted a simple/rough benchmark to which is more > expensive. I tested on a Intel PIII (450MHz 384MB ram) box running Win Xp, > Apache 1.3.26 and PHP 4.2.1, and mysql 3.23.49 and freeBSD of similar stats > (1000MHz, 1G ram). I used the adodb database abstraction layer to make my > connections (which adds extra weigt to the db initialization and queries, > but this is the default method I use to access databases) to a db, and then > queried a smallish db with a "select * from table." I then benchmarked a > file read of a similarily sized file. > > Win DB results average (not including the include of the adodb class): > time index ex time > % > Start 1024676092.32095600 - 0.00% > init db 1024676092.34258300 0.021627 75.19% > query 1024676092.34942600 0.006843 23.79% > close 1024676092.34963100 0.000205 0.71% > Stop 1024676092.34971900 0.000088 0.31% > total - 0.028763 > 100.00% > > Win Filesystem results average: > time index ex time > % > Start 1024676092.35610400 - 0.00% > file open 1024676092.35685300 0.000749 28.59% > read 1024676092.35846200 0.001609 61.41% > close 1024676092.35863700 0.000175 6.68% > Stop 1024676092.35872400 0.000087 3.32% > total - 0.002620 > 100.00% > > freeBSD DB results average (not including the include of the adodb class): > time index ex time > % > Start 1024677559.22131200 - 0.00% > init adodb 1024677559.22266700 0.001355 75.66% > query 1024677559.22303400 0.000367 20.49% > close 1024677559.22307900 0.000045 2.51% > Stop 1024677559.22310300 0.000024 1.34% > total - 0.001791 > 100.00% > > freeBSD Filesystem results average: > time index ex time > % > Start 1024677559.22374400 - > 0.00% > file open 1024677559.22380700 0.000063 11.23% > read 1024677559.22423200 0.000425 75.76% > close 1024677559.22428200 0.000050 8.91% > Stop 1024677559.22430500 0.000023 4.10% > total - 0.000561 > 100.00% > > On the win box, file system access was 11 times faster, while on the freeBSD > box, file system access was 3 times faster. The include of the adodb class > is not benchmarked, as part of this test, that that adds extra overhead as > well. > > I suppose that filesystem access is faster. > > Michael > > "Erik Price" <[EMAIL PROTECTED]> wrote in message > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > > > On Friday, June 21, 2002, at 11:19 AM, mike wrote: > > > > > I was reading somewhere (can't remember where) that connecting to a db > > > is a > > > pretty costly transaction. DB queries aside, does anyone know of any > > > benchmarks that demonstrate file access vs. db connections? > > > > > > Similarily, while DB queries offer alot of power, would it be cheaper > > > (faster) to drop simple information that does not require heavy queries > > > into > > > a file and access it through the file system? > > > > I don't have any stats, but I think it really depends. If you're > > executing a really complex query that uses like six JOINs and eight > > WHERE clauses, then the bottleneck is the DB and not the DB access > > itself, so it would probably be quicker to have this information ready > > in a file (or even better, cached in memory somehow, though I have no > > experience doing this). But I believe that with a simpler DB query, a > > DB access is faster than a file read. > > > > Here's something that turned up in Google... > > http://phplens.com/lens/php-book/optimizing-debugging-php.php > > > > > > Erik > > > > > > > > > > ---- > > > > Erik Price > > Web Developer Temp > > Media Lab, H.H. Brown > > [EMAIL PROTECTED] > >
Mike, I'm not quite sure what you are trying to achieve, but if holding the data in a file is realistically an option i.e. your data is static, then why not consider holding your final output e.g. your web page/partpage in the file system? If you need your php script to generate it in the first place or regenerate it on request there are simple techniques to allow you to do this without reassembling it on every request. Basically you get your script to see if the output has been already created (if (file_exist), and simply redirect or include the output if it does. If it isn't you can get the script to run on and produce the output with output buffering (ob_start etc), switched on. Then at the end of the script you save off the output buffer to the file. In any case, I strongly advise you try your test without the adodb wrapper. You may be suprised just how much overhead it creates. George -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
