All the Perl that's Practical to Extract and Report

Five Great Technologies For 2004

davorg on 2003-12-03T15:03:21

Computer Weekly lists its five technologies to watch for next year. The list includes Open Source Software (again).

2004, huh?

brian_d_foy on 2003-12-03T18:57:19

I guess their editorial cycle is really long. For instance, hasn't "utility computing" been around as long as the computer (and I don't just mean the electronic ones)? They say it is "Pay only for the processing power you use."

Well, back in the Day, when DEC was strong and I was using VMS at government labs, I saw accounting numbers for how many cycles I used and how much that would cost our group.

This is just another example of my notion, which I hope to turn into an idea, that so-called utility computing is just one side off the swing of a pendulum. This pendulum ossicilates between people who want to control everything on their desktop (so they can install, run, whatever anything they want), and people who want someone else (MIS, IT, geeks down the hall) to manage it for them.

Often times, these two groups of people are the same. They have everything on their desktop so they do all their own support, then they realize that they are doing all their own support, so they get someone else to do it. After a while, and by my unscientific observations, this time is 5 to 7 years, they realize they cannot do everything they want because someone else controls the computers, so they bring it back to the desktop.

Why are things like ColdFusion and PHP so popular now? Because Perl was popular before. It will come around again, maybe with a different language or technology, but people swing back and forth between flexibility and ease. How long ago was COBOL created?

Now, the real trick is to time your business cycle to know when the oscillation is going to happen (i.e. figure out its period). Different places are slightly out of phase, and some places even have different phases in the same company. Big companies may have different departments at different places in the flexibility/ease oscillation.

But, figuring out when this is going to happen is clouded by the fact that hack "computer journalists" think every development is a new thing. Ask people who have been around a long time, or someone like Ziggy who has an encyclopedic knowledge of the history of computing starting somewhere around the invention of the papyrus scroll, and you will see that almost nothing is new. Indeed, this is one thing that James Burke's Knowledge Web could elucidate.

If you beleive the hype, you will end up chasing the next "new" thing just in time to catch it and start chasing the new next "new" thing. Indeed, I think the article looks like jjohn's MarkovBlogger than a serious article on what's happening in 2004. :)

Re:2004, huh?

Ovid on 2003-12-03T21:56:27

Great reply. I only have one teeny side comment:

Why are things like ColdFusion and PHP so popular now? Because Perl was popular before. It will come around again, maybe with a different language or technology, but people swing back and forth between flexibility and ease. How long ago was COBOL created?

I think the example of COBOL weakens your argument. Consider the era in which it was created. New language technologies weren't coming around the way that they are now and COBOL had no serious competition. As a result, we have many billions of dollars worth of COBOL code strewn around the world and when a company is looking at their multi-million dollar COBOL code base and thinking about migration, they think "large projects fail" (and they're right). Further, the business rules embedded in decades of COBOL code are mostly known by older programmers, many of whom aren't going to look at Java or Perl and really understand what's going on. It's like being used to working on a Ford Model T and being handed a Ford Taurus: you'll likely recognize what's under the hood, but you can't even do a tune-up.

In other words, COBOL hangs on because regardless of whatever business cycles were taking place in the sixties and seventies, choosing anything but COBOL was not part of that cycle and now it's often deemed too expensive to replace. Today, the choice of COBOL is not subject to cycles the way that the choice of ColdFusion and PHP are, but the article was speculating about evolutionary trends (a model COBOL fits) rather than cyclical ones (which ColdFusion and PHP fit).

Re:2004, huh?

brian_d_foy on 2003-12-03T22:13:57

I was not trying to make a comment on COBOL itself, but rather the notion that a natural language, executive friendly language was the way to go. Now that I think of it, that's SQL too.

TCO

Ovid on 2003-12-03T19:49:56

Interesting quote listed in the article:

People are voting with their hearts, not their heads. The total cost of ownership of open source is open to question. It is a bit like the move from mainframe to server-based computing: it may cost less to buy, but in the long-term, it may cost more to manage and maintain.

This is the bugaboo that open-source software has to overcome, but it frequently succeeds. From personal experience, I can tell you that I would much rather administer Apache than IIS. IIS has this annoying habit of "binding" functionality in such a way that changing one setting can silently reset others. No such problem with Apache.

That illustrates why I think the TCO argument is flawed: these products will get cheaper to manage and maintain when enough people use them and they have a chance to mature. If people refuse to take risks because something is unknown, it will probably remain unknown and immature. Windows wouldn't be so easy to use if no one had used it.

Re:TCO

gav on 2003-12-03T20:39:07

The "cheaper to manage and maintain" argument is a bad one. I hate to use the IIS vs Apache example, but I will. You could argue IIS is cheaper to manage because it has a pretty GUI that an unskilled (and thus lower salaried) person could figure out, with Apache you need to hire somebody that knows what they're doing (though they could manage more servers). This ignores the amount of money that you lose when your network goes down when the next Code Red/Nimda/etc hits.

Re:TCO

brian_d_foy on 2003-12-03T20:46:32

That is a bit simplistic though. I have seen a lot of IIS installations. They take a couple hundred Dell rack-mountables to keep everything moving. I think the numbers would come out very different in a full computation.

Re:TCO

Dom2 on 2003-12-03T21:30:15

My favorite experience with IIS was the day I had to change a setting all 255 virtualhosts that the box was supporting. I just laughed. It would have taken two days to do all the clicking around in that godforsaken admin tool. So naturally, I delegated. :-)

-Dom

Why's it cost?

petdance on 2003-12-04T16:19:00

I'm dismayed that all discussion of open source revolved around the monetary costs of purchase and TCO, and not at all on the less tangible costs of having software that you're powerless to modify. Even if MS dropped prices, even to zero, they'd still have a big battle against free software.