On Tue, Jul 30, 2002 at 01:05:18AM +0000, Jeremy Hiatt wrote: > Hi, > > We’re planning on buying more hardware soon and have been pondering the > machine configuration mentioned in Rasmus Lerdorf’s “Programming PHP” book > (O’Reilly): a squid cache redirector sending traffic to multiple Apache > servers each running a MySQL slave which is replicated from a master MySQL > server. It would look something like: > > Squid cache redirector > | > Apache1 Apache2 Apache3 > MySQL slave MySQL slave MySQL slave > | > Master MySQL server
Yes, it works well. Good choice. > Although we don’t yet make use of it, our code has been optimized to send > writes to a master database and keep the reads local. We’re at about 99.92% > read key efficiency with our MySQL/Apache/PHP application. Up until now > we’ve been pseudo-scaling by adding dual 1.0 GHz p3 machines with 1 GB > memory and running both apache + mysql on them without replication. > Although this has worked for us, it has not allowed us to accommodate many > users at once on any single server. Makes sense. > What hardware configuration and vendors would you suggest for a database > server (MySQL) accommodating upwards of 100,000 users at once? So far we’ve > used only Intel Pentium machines and thusly we’re hesitant to make a leap in > architecture. Right now we don’t have the funds to make a mistake in buying > a larger, “corporate” solution so we’re doing a lot of reading and watching. > If you were to follow the schema above, what hardware would you choose? Can you translate 100,000 users into database numbers? How many SELECTs per second, UPDATEs per second, and so on? That'd help a lot. > Another question: would a mysql master doing writes only and replicating to > multiple servers work better with more clock speed or more cache i.e. would > a dual 700Mhz Xeon with 1-2Mb cache (more cache) work better than a 1.4Ghz > Xeon with fixed 512k cache (more clock cycles)? I have also heard mixed > opinions about a performance loss with >2 Xeon processors-- can anyone > confirm this? I suspect that you'll get more bang for the buck out of memory than CPU power. Once you're above 1GHz of power, the CPU ceases to be the bottleneck unless you're really pumping a lot of data... Jeremy -- Jeremy D. Zawodny | Perl, Web, MySQL, Linux Magazine, Yahoo! <[EMAIL PROTECTED]> | http://jeremy.zawodny.com/ MySQL 3.23.51: up 61 days, processed 1,270,149,448 queries (240/sec. avg) --------------------------------------------------------------------- Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail <[EMAIL PROTECTED]> To unsubscribe, e-mail <[EMAIL PROTECTED]> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
