Re: oracle has customers over a barrel
Date: Sat, 12 Sep 2009 12:38:51 +0000 (UTC)
On Fri, 11 Sep 2009 14:14:39 -0700, joel garry wrote:
> I'd propose a more radical solution: all the new application paradigms
> are _not databases_. They need a different name. Heap has another
> meaning - how about piles? The reference to hemorrhoids is probably
And none of the new application paradigms are as successful as is the database paradigm. To make things very simple: SQL is still very much in line with business needs. You can think about the records or you can think about subsets and relations. Set theory is, in my humble opinion, still the best fit for the general business needs. None of the know categories in mathematics is a better fit: vector spaces, differentiable manifolds, Abelian groups, Lie algebras or complete metric spaces are all very significant mathematical categories, in the sense of category theory, but none of them is even remotely as general as the "naive set theory" (official name for the set theory minus the axiom of choice, Zorn's lemma and the well-ordering theorem). The problem of programming paradigm is the problem of mapping the real world problems into theoretical model that is consistent and allows an easy manipulation. The naive set theory is still by far the best fit for the job.
The whole controversy about the databases stems from the fact that they strive to solve a very general problem: how to quicky extract a subset of data from the possibly huge lot. The problem has 3 layers:
- One has to be able to describe the desired subset in very precise terms.
- One has to able to execute this extraction in a reasonable time.
- The application must be able to deal with the entire subset and not just records.
Unfortunately, all 3 layers require a considerable amount of knowledge and understanding and rather expensive software. There is an ongoing rethinking effort, throughout the industry and in my opinion, that effort was started by ever increasing prices of Oracle RDBMS. Essentially, all databases that do not have special uptime requirements are being moved off to other databases, of open source variety, if possible. In essence, there is no need to keep HR data, resumes, personal information and alike in an Oracle RDBMS. That info is used from Monday to Friday, from 9AM - 5PM, period. Data model is very simple and well known. Good enough application can be quickly manufactured by using Symfony, Zend Framework, Django, Ruby on Rails or something like that. Postgres or MySQL are good enough. The same thing applies to salaries, working hours and office supplies.
Only the most critical, client-facing databases are important enough to warrant the purchase of expensive Oracle licenses. The whole thing doesn't just stop with the databases, it goes much deeper. Not only is Oracle RDBMS being replaced by PostgreSQL, WebLogic is being replaced by JBoss, Tomcat, Hibernate and Maven, there are pilot projects investigating Django, Ruby on Rails and Symfony all over the place. The company that I work for offers a choice of Ubuntu or Windows Vista on desktops, with the strong preference for Ubuntu inside the IT and the strong preference for Windows outside the IT. Some servers that do not need special handling, used to run the internal Apache servers, Samba and NFS servers are all moved from Red Hat to CentOS.
In the long run, I believe that this movement toward cheaper open source software is positive and that employers are willing to invest more into the highly qualified individuals with strong expertize in databases, scritpting languages and OSS operating systems then into licensing proprietary software like Oracle. There is a strong feeling that companies are being extorted. That is why I made a significant effort to learn Perl and PHP and to manage performance on Linux variants. Needless to say, when presented with alternatives, I chose Ubuntu.
As someone has said, the economic paradigm in a crisis is the paradigm of "good enough". The biggest problem for Oracle Corp. is the fact that the OSS databases like PostgreSQL are good enough and much cheaper than Oracle RDBMS. PostgreSQL can do everything that Oracle 8i was capable of, at the fraction of the cost. Oracle Corp. will have to adjust or suffer the consequences. If my memory serves me right, once upon a time there was a great computer company, 2nd only to the mighty IBM, with an excellent engineering, probably the best OS that the computing world has ever seen and very advanced hardware architectures. The company was based at the Mill, in Maynard, MA and it was called DEC (Digital Equipment Corporation). All of it was brouhgt down by the incompetent sales and marketing policies comparable to those of Al Capone. The very same things can be said about the Oracle Corp. Dinosaurs are big and powerful but notoriously hard to adjust to the changed circumstances. DEC was the dinosaur that went extinct during the savings and loan crisis in the early '90s. Unfortunately, every company strives to become a huge and powerful dinosaur. These days, a meteor has struck again. The fate of the Oracle Corp. is far from certain. The fate of the database paradigm as such is, in my opinion, not in danger.Received on Sat Sep 12 2009 - 07:38:51 CDT