Re: Need your help [arguments for open source databases]

• Previous message: Gary Stainburn: "Re: HELP!!"
• In reply to: Willem van der Walt: "Re: Need your help [arguments for open source databases]"
• Next in thread: Reuben D. Budiardja: "Re: Need your help [arguments for open source databases]"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

To: redhat-list@redhat.com
Date: Thu, 24 Jul 2003 07:07:04 -0700 (PDT)

> > I would like to have your opinions about open source databases versus
> > commercial ones. I had a _serious_ debate with some of my colleagues
> > about open source in general but when I started to compare mysql to
> > oracle, they literally fired at me as if I had made a blasphemy... They
> > are stating that Oracle is above any comparison and that it is the must
> > in the universe of relational databases.

As well they should. There's nothing wrong with open-source databases,
it's just that MySQL doesn't qualify. It doesn't fulfill hardly ANY of
the requirements of a relational database, like views, updatable views,
subselects, triggers, custom functions/procedures, and other such things.
Although it has finally started supporting transactions, I don't know if
it supports fully serializable transactions. I would suggest before
making recommendations on databases, you pick up one of the suggested
books on http://www.dbdebunk.com/

Open Source databases are great. Just choose a real one like PostgreSQL
or SAP-DB. Firebird may be a decent one, but I don't know much about it.

> > Since, I am searching some objective informations in order to classify
> > these databases. I found (and was surprise to find) that the real
> > alternative to oracle were rather postgresql than mysql: when comparing
> > the features.

Don't forget SAP-DB.

The main difference between PostgreSQL and Oracle is manageability. With
Oracle, I can split up my table where all of the values where id>30000 are
on disk A, and all of the values wher id<=30000 are on disk B, thus
allowing full table scans to go at double speed. With Postgres, the best
you can do is store different tables on different disks, and even that is
a bit hacky.

The nice thing about Postgres is that it is much easier to tune. After
you get done changing the hideously low default values in their
configuration file, it does pretty well itself. Using the "explain"
command you can even have the optimizer tell you why it performed a
certain query a certain way.

Note that the defaults on Postgres are set for 80s-era machines - so that
it will work out-of-the-box for all configurations. You should ALWAYS
modify those values before doing tests.

Anyway, PostgreSQL is VERY quick and VERY snappy. With a bunch of work,
you can even get some replication on it.

PostgreSQL also has a really nice rewriting system which allows for very
intricate updateable views. I do not know if Oracle has as good of a
system here, although I think Oracle is SQLXX compliant while PostgreSQL
is not.

As to benchmarks, Oracle doesn't allow the release of benchmarks without
the approval of the marketing department. Doing otherwise is a violation
of their licensing agreements, and they can sue your pants off. Likewise
with SQL Server.

Now, if you want Oracle's manageability, you can use SAP-DB. In fact, it
has an Oracle 7 compatibility mode, which allows you to basically use your
Oracle DBAs for MySQL stuff. However, SAP-DB is terrible to set up.

Oracle also has the ability to link tables on separate database systems.
PostgreSQL does not have this, but I'm not sure about SAP-DB.

Personally, I would use PostgreSQL any day of the week, because I'm not a
big fan of paying full-time DBAs, and Postgres, although has fewer
management features, more or less self-optimizes. However, for extremely
intensive workloads, Oracle would probably be a better choice. For
example, the travel industry is probably well off that they use Oracle
over PostgreSQL, but most of the rest of us are probably better off with
PostgreSQL.

Jon

> >
> > But I didn't find any head to head comparison between open
> > source/commercial databases.
> >
> > Do you know any benchmark results that made this kind of comparison?
> >
> > Thanks a lot for any suggestion.
> >
> > And again, my apologizes if this is not the right place to ask such
> > questions.
> >
> > Best regards,
> > Kader.
> >
> >
> >
>
>
> --
> redhat-list mailing list
> unsubscribe mailto:redhat-list-request@redhat.com?subject=unsubscribe
> https://www.redhat.com/mailman/listinfo/redhat-list
>

-- 
redhat-list mailing list
unsubscribe mailto:redhat-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list

• Previous message: Gary Stainburn: "Re: HELP!!"
• In reply to: Willem van der Walt: "Re: Need your help [arguments for open source databases]"
• Next in thread: Reuben D. Budiardja: "Re: Need your help [arguments for open source databases]"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]