The One True Database Engine

We need more tolerance in the world of databases.

In the computer world (and, for reasons that will become clear below, I include in this world all of the wonderful computer-based gadgets that are appearing), people tend to have strong opinions.

The problem with polarised opinions is that they are typically founded more on prejudice than information – just like teenagers and trainers. When a teenager is buying a new pair of trainers, they don’t try on every pair in the shop asking “Does my bum look big in these?”; they try on every model of one brand in the shop.

Which brand? Well, that depends on the teenager. The point is that any given teenager already knows that Nike (or Reebok, or Adidas, or Puma, whatever) is the one true brand. There is no other. Every other make is demonstrably the work of the devil and will instantly destroy their street cred, even if they try them on in a darkened room with only parents present.

And so it is with database people – they know the name of the one true database engine (OTDE) and must attack, denigrate and belittle any false database engines. Yeah, even as it is written in Books-On-Line.

Given my age I have no idea how a teenager chooses the one-true-trainer brand but I do know how most database people become acolytes.

Distressingly most database people never even get to choose the OTDE. They start working for a company which is an Oracle/DB2/SQL Server shop. They gradually learn the skills necessary to drive it. After a month they have somehow magically been transformed from a normal reasonable person into a complete database bigot.

The sad fact is that we end loving the one in which we have invested time learning to love. (OK, this is the point where the whole teenager/trainer analogy falls down – teenagers do not usually spend months learning how to tie the laces of Nike trainers, but I think I’ve got all the mileage I wanted out of that idea anyway).

Now, let’s face it, most database people are geeks. I know I am. And the odd thing is that the behaviour described above is non-typical for geeks when they make other choices. There the process is very different (but equally bizarre and logically unsupportable).

Every now and then a totally new gadget appears (think about PDAs, Mobile Phones, memory sticks, iPods etc.). The geek is instantly attracted to the new toy; obviously. He (or she, but mainly he I guess) will gleefully devour all of the technical specs. Which has more memory, which runs fastest, which has the greatest number of features?

Since technical products are highly competitive, the specs. are often very close. Or, more accurately, the balance of specs. is very close. This one has more memory but this one has a better processor. This one is more fragile but it is lighter as a result.

It may take days or even weeks of study for the decision to be made. Up until that point, the scales are finely balanced. But once that final decision is made, once the money has been exchanged for the shiny toy, all of that instantly changes. The alternative products are, quite clearly rubbish; total junk. There is only one-true…..whatever.

Well, the point of all this is that database engines are like gadgets; they all have a complex set of specs. which tend to balance out. This database engine has the better query optimiser but punishes the CPU; that engine is more difficult to drive but uses less memory.

Then there is the question of non-technical issues which most techies pretend are irrelevant but are often, in truth, over-riding. These include the marketing clout of the company involved, the level of support, the cost and so on. I would argue that these are as important as the technical issues, so we end up with a very complex matrix of pros and cons for each engine.

However, when all of these are taken into account it is clear that there cannot, really, be OTDE that is far, far better than the opposition in all and every way; for the simple reason that market forces are there, in the background, quietly killing off the less fit.

It is easy to forget, in the present climate, that the savannah of the computing world was once roamed by many species of database engine that are now essentially extinct. If you don’t believe me, stop reading now and try to name all of the PC-based, non client-server, database engines that you can.

Back in 1992, for that particular segment of the database market, I wrote a bunch-test review that covered 18 different database engines:
Paradox
Clipper
dBASE
Matt Whelan's Compiler Kit
DataEase FoxPro
Delta V
Dataflex
R:BASE
ACCESS
FoxPro for Windows
dBFast
Omnis 7
SKYLIGHT
WindowBase
APPROACH
Superbase
DataEase for Windows

How many of these remain as a significant force in the world of databases? (I’m aware that of aficionados of many of these will argue “Well, is certainly still a major player!”. True, and all 34 acolytes would agree (35 if you count the designer’s Mum.)

And, no, that last paragraph doesn’t mean that I hate the products listed above. I thought Skylight was great as a simple database engine and once gave it an editor’s award for best product in a category.

In addition I have, in the past, done a great deal of development work using both dBASE and Paradox and loved them both for the strengths they had. But over time, some of the engine evolved and got better, others didn’t. And unnatural selection took out the ones which didn’t evolve.

The bottom line is that all of the database engines that the majority of computing people can name off the top of their heads (DB2, Oracle, SQL Server, MySQL, Access) are all perfectly serviceable engines. None of them are the work of the devil. All of them have significant strengths. Oracle people and DB2 people can share a drink and even opinions, without violence. I know it can be done, I have actually seen it happen at a database conference. All that is required is a little tolerance.