[CentOS] mysql and windows

Sat Sep 29 00:02:38 UTC 2007
gjgowey at tmo.blackberry.net <gjgowey at tmo.blackberry.net>

My problem with mysql just stems from watching how it has developed from the early days.  It has a sleepycat core (nothing wrong with that), but the added functionality to make it an rdbms revolves around whatever is popular at the time and quick to do rather than what do other databases do.

I'm dating myself, but I remember when the mysql crowd used to scream if you mentioned transactions.  They used to have a full page on their site about how transactions were unnecessary and getting the data in as fast as possible was a better idea.  

Now? When did they put transactions in?  2-3 Years ago?  A feature that all other SQL based rdbms' including postgres have had since the beginning of their existence.  There are other outstanding technical reasons that make me stay away from mysql whenever possible.  I just have severe disagreements with their development ideology which ultimately results in a flawed end product.


Sent from my BlackBerry wireless handheld.

-----Original Message-----
From: Peter Arremann <loony at loonybin.org>

Date: Fri, 28 Sep 2007 19:52:25 
To:CentOS mailing list <centos at centos.org>
Subject: Re: [CentOS] mysql and windows

On Friday 28 September 2007, Miark wrote:
> On Fri, 28 Sep 2007 21:01:07 +0000, gjgowey at tmo.blackberry.net
> wrote:
> > Mysql is to databases what Lincoln logs are to cinder blocks.
> What open source DB progs do you like--if any?

You shouldn't have asked that question... :)

When the dust settles at the end of the religious war you just started, you'll
usually end up with people mostly agreeing on the following points:

* mysql is faster in high volume, mostly read, simple query scenarios.
Especially if your app can use a query cache, its not even close. pg wins
when you have much more complex queries or more writes.

* pg scales higher - I just finished some testing on a 32 core box and it
scaled almost as well as oracle (oracle got 23 times single core performance,
pg 21 times). mysql levels off pretty much at 4-8 (depending on workload) -
larger machines often get slower.This is now changing with falcon though, so
look out for 5.2...

* pg allows higher concurrency. If you pull data, then mangle it for a while
before going back to the DB with the next query, mysql can sometimes become
dog slow... pg with row versioning allows for higher concurrency. Again,
falcon fixes a lot of that.

* pg recovers better than mysql. However if recovery fails for some reason,
finding someone who can salvage at least something, is next to impossible.

* backup is faster in pg, recover is faster in mysql...

* pg adheres much better to the sql standards than mysql

In the end, this is enough reason for me to go with pg in most cases - and if
that is too large, sqlite... Between the two, there is little space left for
mysql in my world.

CentOS mailing list
CentOS at centos.org