Re: [GENERAL] Re: MySQL / PostgreSQL (was: Postgres object orientation)
Ronald Chmara <ron@opus1.com>
From: Ron Chmara <ron@Opus1.COM>
To: Alain TESIO <tesio@easynet.fr>
Cc: pgsql-general@hub.org, Guillaume.Rousse@univ-reunion.fr
Date: 2000-01-08T03:17:30Z
Lists: pgsql-bugs
Alain TESIO wrote: > > Subsidiary question : why is mySQL excluded from RDBMS > > comparison on postgress www site ? I believe it's fairly stated: mySQL is not an RDBMS, so it is not listed. A similar question: Why is Filemaker excluded? FoxPro? They all have a speedy, flat file, structure, glued together in such a way that relational-like interactions are available. > Maybe because it's much faster ;) The speed of not having some relational feature that you desparately need, but is not availabe, is approximately zero. :-) I like it for simple structures, simple db's, but as soon as you're trying to manage 30-40 tables, mySQL starts to get in its own way. For running SQL to a flat db? It's much better than PostgreSQL. 200 Tables? Don't even think about it, get an RDBMS. > Sure, MySQL doesn't support transactions, rollbacks, ... Foreign keys... > I've compared both engines and MySQL is much faster. Yup. Different design goals. > However I'll need transaction to ensure reliability > for the database updates. I've thought at using PostgreSQL > for updates, and MySQL for select, the database being > dumped from PostgreSQL and reloaded into MySQL every > night. Probably with specific queries and scripts rather > than a dump to get a MySQL-compliant dump file. > Has anyone an experience about a similar solution ? We are in the midst of nightly dumps of mySQL -> PostgreSQL, in order to gradually migrate to a more robust solution... however, it seems like your proposed idea _could_ work, with some decent scripting. Fast for simple scans, too... basically, you'd need text dumps, which are then re-wrapped to mySQL-friendly insert (or whatever) statements. HTH, -Bop