Open Source

Why the Oracle Application Stack should not happen


Apparently, Oracle is seriously considering putting together its own full application stack. There seems to be a lot of debate about this, both positive and negative. Personally, I think I will have to side with the naysayers on this one.

First of all, the last thing the world needs is more confusion amongst the Linux distributions. I understand that Oracle is looking to purchase an existing distribution, not start their own. But do open source projects really manage well after being purchased by a large corporation? SuSE seems to be a bit unhealthy after being bought by Novell. Granted, Novell has "Fido's magic touch" where everything they touch turns to dog doo. WordPerfect. Corel. Their own products. And so on. Much of the open source community seems to be "personality driven." The movement of one or two key contributors to a project can cripple it. Not just because they were cornerstones of development, but if they leave, so do many other people. Forking is another common problem after an open source project gets purchased. This is actually the main reasons why I use BSD instead of Linux; I feel that there is too much churn in the Linux community. If  I were a Linux user, my biggest fear would be "what if Linus Torvalds gets hit by a bus?" I can imagine the power scramble if that were to happen, and it scares me. So the idea of Oracle purchasing or starting a Linux distribution of their own would worry me, particularly if they were to purchase one.

Another issue with the idea of Oracle having a stack, is that I am not sure if Oracle is a company I would want to have to deal with. Their website is a frightening place indeed; finding useful information in a usable format can frequently take hours. Just try to find the SELECT syntax on their website, I dare you. I do not think that Oracle understands how to interface with customers on that type of basic level well enough to want to have my OS coming from them.

Another arena where Oracle falls woefully short already is in their management tools. Everything about installing, configuring, and maintaining Oracle's database products is pure misery. Everything is done wrong, as far as I am concerned. They have visual tools such as Oracle Enterprise Manager that simply do not work right. For example, if you put the cursor into a field and start typing, the first character is usually dropped. The interfaces on their visual tools stinks as well. Oracle Enterprise Manager has "features" on the menu that when selected, tell you to use "Oracle Enterprise Manager Console." Isn't that the tool I am currently in? Even the GUI version of SQL Plus is a dog; it has a maximum line length, forcing me to wrap lines by hand, and does not even do my the courtesy of putting a vertical bare showing me where to wrap them. Oracle products do not install correctly; the DLLs needed for ASP and ASP.Net connectivity have the file permissions set incorrectly, a problem that has persisted through a number of major revisions. Even trying to get clients talking to Oracle is a pain. Oracle needs multiple hundreds of megabytes worth of garbage to have a desktop talking to an Oracle database, whereas MySQL and Microsoft SQL Server just need a tiny ODBC (or JDBC, or whatever the right method is for your purpose) driver. Overall, the last thing I want is for my OS to be delivered by a company with this mentality.

Oracle is also very, very bad about delivering patches, and they do not seem to have a handle on security. In terms of timely patch releases, they make Microsoft look like perfection. Oracle has a bad habit of outright ignoring critical security flaws for months or years at a time, even after being told about them. Their patch cycle seems to be quarterly; meanwhile, Microsoft gets criticized for monthly patches. Oracle also does not seem to understand automatic patching. Again, these are traits I simply do not want in the source of my OS.

As it stands now, a good portion of Oracle's stack is not even their own software. Oracle Application Server appears to be a hacked up version of Apache. They do not have any languages of their own (or even re-packaged/re-branded) outside of PL/SQL which you will not be writing applications in. Right now, the only part of the LAMP stack that Oracle can play a role in is the M. You can have a LAOP stack if you want. Oracle is considering grabbing the L. They are still missing the A and P. Indeed, when one looks at what Oracle does well (high performance, scalable database server), I would much rather prefer that Oracle go after the A and not the L! Web servers are more closely related to databases, in terms of how they get written. You simply do not try to start from the middle of the stack and work your way out like Oracle is considering. You need to start from one end or the other and go across. Red Hat had the L, they bought JBoss to get the A. They still need M and P. Oracle is trying to start at M and then go for the L. This just does not work. This is not a stack, this is patchwork insanity.

The idea behind a stack is that you have a group of tools that sit upon each other and play nicely with each other. Patchwork stacks just don't cut it. Components within a stack are often not best of breed by themselves, but the combination works great. Look at LAMP: the P is not so great (Perl is poor for web development, PHP is wretched in general, Python just isn't very popular), and MySQL is not quite top-tier yet (although it is still great). But LAMP works great. Oracle does not know how to make their software play nice with other software. The idea of them trying to build a stack is laughable, at best.

J.Ja

About

Justin James is the Lead Architect for Conigent.

0 comments

Editor's Picks