More on SQLite vs MySQL vs PostgreSQL

Matts on 2003-01-28T16:08:18

D. Hipp sent me his benchmarking code, so I ran it on my optimised Pg box. The optimizations are really just giving Pg some real memory to chew on. This box has 4G of ram, and Pg out of the box is optimized for a box with 8M of ram (no, really!). My config is here if you're interested.

Results of my testing.

One thing to note is that the tests use the sqlite (and psql) command line tool, whereas a better test for perl hackers would be one using DBI. But this thing is written in tcl, and I don't want to hack on it any longer than absolutely necessary! ;-)

---

Hmm...

pdcawley on 2003-01-28T18:52:24

I notice that none of the tests involve joins, or concurrency, which is where I'd expect PostgreSQL to start to pull ahead.

Re:Hmm...

Matts on 2003-01-28T19:48:26

I agree. Not my tests though - I just wanted to show that pg can do better than the original ;-)

nosync??

jcavanaugh on 2003-01-28T21:54:39

How is this configured for sqlite??

--John Cavanaugh

Re:nosync??

Matts on 2003-01-28T23:08:40

It's a pragma. See the docs on www.sqlite.org.

results 404

jnoble on 2003-01-31T14:06:56

The link to your results is 404 -- can you put it back for a while? Thanks!

Re:results 404

Matts on 2003-01-31T16:34:41

Oh, I moved everything. Will post a new top level journal entry about it.