I have three machines which I'm working with a specific database on. One is a development server, the second a production server, and the third a testing server for when code may cause the development server to crash (which is the development server for numerous applications and sites so this is rather disruptive).
The production server is a Xeon 2.4GHz with 1GB of DDR RAM and 512MB of swap. It is SCSI based and has plenty of diskspace free (~50%) for /var/lib/mysql databases. There are typically around 90 active databases on the server. It's running a RedHat Linux kernel 2.4.7-10. The development server is an Intel ix86 Pentium III based 1GHz with 768MB of ECC SDRAM and 256MB of swap. It is also SCSI based but only has around 20% free diskspace. There are 100+ databases on the server but only about 95 are active and in use. Since this machine doesn't have nearly the httpd load on it, mysql queries are significantly less as well. It's running a RedHat Linux kernel 2.4.7-10 as well. The testing server is an old Pentium 266MMX machine overclocked to 400MHz with 768MB of RAM/Swap and is IDE based with plenty of diskspace (since it's a great deal cheaper). There is typically only two to three databases on this server at any given time undergoing various degrees of testing. This machine runs debian with linux kernel 2.4.24 and served previously in life as an NT workstation. All three machines are runnign identicle versions of MySQL (4.0.17), Apache (1.3.29), and PHP (4.3.4). The weird thing is that a dump of a small mid-sized database (~25MB - 140,000 records) from any of the three machines and imported into another via CLI will take a very long time (45+ minutes) on the two higher end servers, but takes a very short time (~15 minutes) on the low end testing server. Validating our raw data (in files) against the database typically takes less time on our development and production servers but not significantly enough compared to the difference in computing power between the servers and the workstation. Thinking it may have simply been the load on the high end servers I tested the development server in the middle of the night when no one was on it and came out with the same results. Now the testing server performs poorly when two or three tests are being done on different databases at the same time whereas on the development and production servers they run at only a slightly degraded state except for the occassional bug which freezes up the servers after using up all memory and swap. Will the number of databases being handled by MySQL affect it's speed regardless of the actual load being put on MySQL or should I look farther into OS issues which could be causing system hangups and slowdowns? The age of the OSes in the high end servers is a bit of a sore point but cannot be addressed at this point in time and are currently as secure as possible in all relevant software interacting with the outside world. ---- Dan Brown -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
