Home Messages Index
[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index

Re: Is the SQL database market dead?

__/ [ mlw ] on Tuesday 21 March 2006 17:15 \__

> I read in slashdot that Oracle's database market has remained very flat.
> They also note that open source may be, at least, partially responsible.

Based on the reading I have done, Oracle's revenue in the last profit soared
while their stock plummeted.

A couple of days ago, many waves hit the media, which reported Sony's Oracle
defection. This could catalyse a 'cattle effect' that will ultimatele drive
more companies to Open Source substitutes, before Larry can eliminate them
(he even had a go at MySQL, but failed).

The following impartial benchmark does not compliment any of the commercial
database vendors.


MySQL is superior to databases from Oracle, IBM, and Microsoft.

> I started thinking that, while open source databases may have some impact,
> I think it is more hard disk and processor technology that has killed the
> market.
> 10 years ago, a 30G storage system cost about $10,000. Now, you can't even
> buy a hard disk that small.
> 10 years ago, disk speeds of 7200 RPM were high end, now it is standard.

I bought a 300G hard-drive (7200 RPM) a fortnight ago. This external
hard-drive (which includes the housing, software, among some more goodies)
cost me around 140 pounds. Sure I could buy a Linux box for less than that,
but I needed more storage capacity.

> Over 10 years, seek times of hard disk, transfer rates, internal caching,
> internal error correction, and predictive caching has all made the disks of
> 10 or more years ago, seem like vacuum tube technology.
> Now, Oracle, specifically, has a great deal of options and tuning for
> various disk technologies and configurations. 10 years ago, the data extent
> layout, volume management, and other optimizations made the difference
> between something that worked, and something that didn't. Oracle allows
> database tables bigger than physical disk and were allowed to spread across
> volumes.

Let us save some of trouble associated with analysing this. Oracle still sell
databased for the same reason that Windows continues to be sold to many

Senior managers, who are only few years ahead of retirement and edge complete
miscomprehension w.r.t. recent technology, are looking at flashy pamphlets,
pondering if they can afford a strategic change so late in their
professional career. They are too reluctant to accept the advice of
young-but-talented, up-to-the minute, tenacious staff; unfortunately it is
not them who make the final decisions.

> These days, you can get a disk pretty much as large as you need. These
> days, the disks, the CPUs, and I/O subsystems are so fast, that most of the
> time, an un-tuned system will probably work as well as ever needed.
> Most of Oracle is best at is pretty much obsolete. Yes, there are high end
> deployments that could probably make use of the tuning features, but as
> disk and I/O technology continue to improve, their necessity become more
> and more marginal.
> The problem of a database has shifted, now all those things that Oracle has
> that made awesome 10 years ago, are getting in the way of admins today.
> PostgreSQL, for instance, has a similar level of SQL functionality and
> flexibility to Oracle, but since it lacks much of the underlying
> complexity, is more quickly improved, and it is comparable in performance
> on the most likely deployments.
> Sure PostgreSQL and MySQL lack the ability to spread tables across volumes,
> but these days, who needs to? Buy a 250G hard disk. Need bigger, buy a few
> and use RAID.
> Sure PostgreSQL doesn't have as many data analysis tools as Oracle, but
> with disks and CPUs as fast as they are, you won't really notice.

Distributability in this particular case does not justify the cost. It's all
about myths and sales jargon that gives inertia to commerical DB's such as
Oracle's server.

> So, maybe the database market is falling to open source, but I think it is
> more because the hard disks, I/O systems, and CPUs are so much better, that
> the returns of Oracle's tuning and administration complexity are
> diminished.

I think Ellison's acquisition of so many Open Source companies provides more
than a clue.

Best wishes,


Roy S. Schestowitz      | Data lacking semantics is currency in an island
http://Schestowitz.com  |    SuSE Linux    ¦     PGP-Key: 0x74572E8E
  7:25pm  up 13 days 12:02,  7 users,  load average: 0.60, 0.73, 0.74
      http://iuron.com - next generation of search paradigms

[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index