DevLoop has just published a
free paper (US mirror) [zipped .pdf]
benchmarking some popular database systems such as MySQL, PostgreSQL, Firebird, DB2, Informix, Sybase SQL Anywhere, on Linux using JDBC. The focus here is on non-real setups where we can analyze the individual aspects that might affect system performance. The code is available (GPL) so that anyone can reproduce the results.
To view parent comment, click here.
To read all comments associated with this story, please click here.
Member since:2007-02-14
I can believe it, yes. Of course there will always be exceptions to any case. But this benchmark is not measuring the best DB solution for a particular application profile, merely the speed of their operation. Be careful with how such benchmarks can be interpreted as conclusions by poor coders.
Member since:2005-07-06
I'm glad to hear that you found a database that fits your needs. The problem with MySQL or rather MySQL AB is their marketing. They tell you they have the fastest DB in the world. In the next sentence they tell you that they now support transactions, referential integrity, stored procedures and triggers. Most people interpret this as the can have all this at the same time. Unfortunately this is not so.
So when I see tests where MyISAM is compared to what you call "real" database engines I get sort of suspicious. It's bad enough that MySQL AB holds back on telling about the difference in functionality, independent tests doing it is even worse. Most people will look at the performance figures not realizing that the performed task is quite different in e.g. MySQL or Postgresql.
It is only when people have all the relevant facts, that they can make an informed decision on what's the best database for them.
Member since:2005-07-06
Your app must have well structured data and a thoughtfully designed schema.
I work with financial data that, for reasons out of my control, is not structured well. I have to make up for the lack of structure by using complex queries to groups, sort and limit the data returned to the app. Even though the data is used mostly readonly, and I have my database configured for MyISAM, MySQL performance is quite a bit worse then Firebird. I have not tested PostgreSQL before. OracleXE in my case performs best.
For 90% of the people developing webapps out there mysql is great. But if your in a position where you need data integrity, or your trying to query large amounts of poorly structured data, and performance is a consideration, then mysql is probably not the right database for you.
Member since:
2005-08-12
Believe it or not, I am doing enterprise application developement which heavily uses MySQL. We are providing large data sets (tens of GBs) to our customers that their application read but (almost) never update these data, so what is better then MySQL here? Free, small, simple to install and maintain. There are many more use-cases for DBs then seems to fit into your head, so try to investigate why and how various enterprises use MySQL *before* judging them. For application I am talking about, switching to twice slower DB just because you decided it is "real" DB is not an option. Don't know about you, but test is very useful for me, because I *do know* what features these DBs support and it is very informative to know how performace compares for various workloads. As of JDBC layer, yes, it can be buggy and make bad image of DB itself, but then again, In java you and I access DBs using these JDBC drivers.