Yep, there's a huge performance difference, especially on more complex queries and such.
In general complex queries have gotten a 10 to 100% performance boost from 7.1 to 7.2, and again from 7.2. to 7.3. Some queries are literally hundreds of times faster (because they were god awful slow in 7.1 or 7.0) while some aren't noticeable faster, since they were already pretty fast.
7.3.3 is now to a point where only the simplest of queries (select * from table where id=1) Are faster in MySQL now. that and aggregates in the wrong place. I.e. for the highest ID in a table, use select id from table order by id [DESC] limit 1 instead of select max(id) from table.
Of course, a default install with no tuning, is gonna be pretty mediocre. Setting the effective cache size and increasing shared buffers is a requirement for fast performance, but requires the proper kernel settings to do so on many OSes.
That said, we're still running our main server on 7.2, but will upgrade soon to 7.3
A nice feature we just found in 7.3 that isn't in 7.2 and before is the ability to use regexes in substr() function.
select substr('Scott Marlowe x6000','[[:digit:]]')
RESULT:
6000
is pretty sweet, especially for some of the apps we have where we can't mangle the string before we put it into a query (i.e. nettracker log analyzer).
There's lots of little things like that. Plus schemas, which allow you to have "databases within databases" so to speak. I.e. you can have one database, and ten users, and they each get their own name space, so that both Joe and Bob can do:
create table babes (name text, measurements text, phone text, id serial);
and not have name collisions since they'll be created in seperate schemas.
Then each can set permissions to allow other folks to view/edit/delete their data as they see fit. This is nice for collaborative database development.
