Hi, I don't have any experience with FreeBSD; however, I use MySQL on SMP servers with Linux.
1. due to MySQL architecture (1 thread per connection) you should not expect any performance inprovement on a dual processor for a SINGLE SQL query compared to an uniprocessor machine. This is what you get for most of your queries (the slight improvement is coming from CPU speed difference). Multiprocessor only helps when you run multiple queries in parrallel. 2. you should have the same kind of behaviour for your long query (slightly faster, not 2 times slower). Before suspecting your compilation flags, are you sure that your /etc/my.cnf is correctly setup ? Have you tried with the same configuration on both servers ? 3. If you want to optimize your query, you should try EXPLAIN command and make sure that your query is properly using indexes. Hope this helps -- Joseph Bueno Jay Blanchard wrote: >Howdy all, > >We have a curious situation here, and I was wondering if any of you have >encountered this before. > >We have MySQL installed on a FreeBSD machine (733mHz, 256Mb RAM). Compiled >from source. >We installed MySQL on another FreeBSD machine (dual 866mHz, 512Mb RAM). >Compiled from source. > >The reason we went to the new machine is that we have a massive database >that for practical purposes needed to live by itself. We moved this database >from the first machine to the second machine. > >Most of the queries run slightly faster on the new machine, but one query in >particular takes twice the time to run on the new machine. The query ; > >select c10.RecordID, count(*) as Quantity, sum(c10.Minutes) as Minutes >from tblClass10 c10 left outer join tblANI a >on c10.FromNumber = a.ExemptPhone >where c10.RecordID = '100101' >and a.ExemptPhone is null >group by c10.RecordID; > >On the first machine runs in approximately 20 minutes, on the second machine >runs in about 40-45 minutes. tblClass10 has 5.5 million records, tblANI has >3100 records. Out of tblClass10, the records matching the RecordID criteria >'100101' is approximately 170,000 (today). > >We are thinking that it has something to do with the flags that were used to >compile MySQL with reguards to dual processor machines. I am also up for any >query optimization that would work too. I have tried INSERT ...SELECT the >100101 records into a new table and then running the query above (replacing >tblClass10 with the new table), with the results being no faster. > >TFAYHIA! > >Jay Blanchard >Applications Development >nii communications, inc. >210-403-9100 x285 > > --------------------------------------------------------------------- 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
