MJ Ray wrote:
"Martyn Ashworth" martyn.ashworth@eag.uk.com writes:
What advantages are there to migrating across to PostgreSQL now ?
A database that works "better", is more "stable and has less "interesting extensions". My opinion only, some people love mySQL and
The clincher, for me, is that PG supplies ACID transaction support with no (or no significant) speed penalty. MySQL has finally gotten round to it, but only in special (crippled) table types.
Migration should be particularly simple, assuming you're not using any of the aforementioned 'interesting extensions', so you have nothing to lose but your locks... (MS uses table locks to partially implement transactions, the rest they suggest you do in software)
I expect we'll have to wait a few generations for Darwin to work and all the mySQL supporters to have gone insane and not had any surviving offspring. (Hint for the hard of thinking: I'm not being entirely serious here.)
Maybe we could start an airline with MySQL-powered traffic control.
The other problem that i can forsee is that apparantly PostgreSQL has an 8k row limit which is not un-workable but would require more time and effort than MySQL for some of my data structures at the present time.
What the hell are you doing if you expect to do more than 8k in a single row? If it's a single large object, you should use the particular support for it (which I admit I'm bad at using). If it's really unavoidable, you can change that limit to something larger by recompiling.
Some People (those who 'picked up database design as they went along'), like to have all the data in a single table. Lots of fields. It does reduce joins... <wince>
Incidentally, Martyn, I make the above comment assuming such a design has been forced upon you, not that you are one of the Some.
I would appreciate any helpful feedback here as the only useful comparison i've found so far seems to be inconclusive to say the least !!
All comparisons but one's own are inconclusive...
In the performance stakes I understand MySQL wins, Postgres gets second place and Oracle comes last. However, the real difference is very small. Oracle is only worth using if you need linear scalability to very high degree.
Alexis