On October 10th, 2007 shingen (not verified) says:
I have experience of running big (millions of lines, tens GBs of data) and small databases with both MySQL and PostgreSQL.
PostgreSQL performs better with out-of-the-box confs than MySQL, but when you tune the configurations for the server it's not so straight forward (depends on how many concurrent users, how much RAM/CPUs available and so on). Also MySQL older versions perform much worse than newer > 5 versions. MyISAM as table type sucks badly as well, unless you have small db and don't want to use transactions. Potgre seems to perform better on Linux than Windows also.
There are about 20-30 confs that I usually tweak to make MySQL work ok. If you do benchmarks without stating the confs you use or table types or types of queries, they don't give that much.
Once by just optimizing the confs I made one of my MySQL dbs perform roughly twice as fast as it was before. If that was done to the benchmark above, who knows, it might be a tie.
Confs, table types, db size, db version... it all matters
I have experience of running big (millions of lines, tens GBs of data) and small databases with both MySQL and PostgreSQL.
PostgreSQL performs better with out-of-the-box confs than MySQL, but when you tune the configurations for the server it's not so straight forward (depends on how many concurrent users, how much RAM/CPUs available and so on). Also MySQL older versions perform much worse than newer > 5 versions. MyISAM as table type sucks badly as well, unless you have small db and don't want to use transactions. Potgre seems to perform better on Linux than Windows also.
There are about 20-30 confs that I usually tweak to make MySQL work ok. If you do benchmarks without stating the confs you use or table types or types of queries, they don't give that much.
Once by just optimizing the confs I made one of my MySQL dbs perform roughly twice as fast as it was before. If that was done to the benchmark above, who knows, it might be a tie.