COBOL and mainframes...

[Tessian]

Despite the work trying to clean that up, that must have been a great "I told you so!" moment in business history with great timing. Stuff like this is easy to write a business impact doc on:

[cost of system] < [cost of your entire business going down for a week]

Fucking companies want us to implement everything that makes us safer and more reliable, but doesn't want to spend a dime on the tools that make that happen.

On the subject of AIX/Linux, well, AIX is old shit, and Linux is just like any other OS. If you don't patch it regularly, things are going to get hacked.

Unfortunately I was only an intern at the time so I didn't get to do the "I toldya so" but my boss did. And AIX isn't old shit, it's still updated regularly, patched, etc. We run all our database servers on it.

[SineSwiper]

LOL. I'm sorry. You should be running a Linux, FreeBSD, or Solaris box, running Oracle or MySQL. Hell, I'd even go for a MS SQL box over that.

Of course, I can't talk, since we use a ASS 400 system for our billing. I fucking hate mainframe old skool technology. Goddamn thing should have gone out with COBOL. Hell, I'm sure the whole thing is strung together in COBOL.

[Lox]

LOL. I'm sorry. You should be running a Linux, FreeBSD, or Solaris box, running Oracle or MySQL. Hell, I'd even go for a MS SQL box over that.

Of course, I can't talk, since we use a ASS 400 system for our billing. I fucking hate mainframe old skool technology. Goddamn thing should have gone out with COBOL. Hell, I'm sure the whole thing is strung together in COBOL.

Ha. You make it sound like COBOL has gone out. UPS has tons of mainframe systems that are running on COBOL because it would be way too expensive to replace them and the rewards would probably be minimal.

[SineSwiper]

Yeah, because we didn't learn any lessons from Y2K, right? Just continue with that 50-year-old code and systems, and continue paying increasingly higher salaries on dinosaur programmers that know that they are as rare as priceless jewels.

Sometimes companies do braindead things because they really didn't run the numbers on how much money they would save on preventing their entire company from going down the drain.

[Lox]

Bah...Y2K was made out to be a bigger deal then it actually was. I think it's a good idea for some selected mainframe apps be converted, but to migrate all of them is a waste of time and money. And honestly, the mainframe stuff is probably 50 times more stable than the new code being written today.

We do move off of the mainframes as it makes sense financially and for the business. In fact, I know of one where we're removing the mainframe altogether, but it just doesn't make sense to do that for many of the applications we run.

[Tessian]

There was an article on Slashdot a few weeks ago-- companies have found it's cheaper to pay to pull a colbol programmer out of retirement than it is to rewrite their applications into a newer language. From what I've been told by people who used to program in the language it's a solid language that just required way too many lines of code to do anything... but once it's done it'll run forever.

Why pay 6 programmers $60k/year to rewrite all my applications which will be a long, painful process that could cause much downtime and no gain when instead I can just pay 1 programmer $80k/year to maintain it all?

[Lox]

I actually know COBOL. It was taught in my undergrad program (primarily as a way to provide jobs to new graduates since companies need COBOL programmers).

You can do some pretty impressive stuff with it and it's great for some of these backend systems. What it boils down to is that even after rewriting the COBOL into something else, it still has to be maintained.

[SineSwiper]

Bah...Y2K was made out to be a bigger deal then it actually was. I think it's a good idea for some selected mainframe apps be converted, but to migrate all of them is a waste of time and money. And honestly, the mainframe stuff is probably 50 times more stable than the new code being written today.

No, Y2K was a big deal. We just happened to be prepared for it. If we pretended it didn't exist, then we would be in a cats/dogs living together scenario.

Why pay 6 programmers $60k/year to rewrite all my applications which will be a long, painful process that could cause much downtime and no gain when instead I can just pay 1 programmer $80k/year to maintain it all?

Probably more than that for the COBOL guy. Also, let's say it took a year to rewrite those applications. (Nevermind that you could probably just buy the software outright.) It would cost you $360K. In five years, you would have been paying $400K for that one COBOL guy.

Not to mention, the risk of the hardware breaking and the fear that the part doesn't even exist any more. Or the efficiencies gained by using a modern program. A 30-year-old piece of software means a piece of software that hasn't been improved in 30 years.

[Tessian]

Some of these programs don't NEED to be improved depending on what they do. And I love that you bring up an ROI because most companies don't give a shit about an ROI that doesn't work out in a year. My boss and I submitted an ROI to switch internet filtering products that would have cost a good deal more the first year in purchasing hardware, but by year 3 would have saved us over $50k; they didn't give a shit. Why do you think companies will go with the cheaper part even though it'll need replacing in a few years and cost more to do so? Cause it makes me look better in this year's budget that's why. So short sighted...

[SineSwiper]

God, I know. Don't these companies understand that long-term goals are the only way to make it long-term?

[Flip]

God, I know. Don't these companies understand that long-term goals are the only way to make it long-term?

Most companies understand this. Wall Street is always looking well ahead, which is why despite the fact that AOL makes a shit ton of money right now, we still get slammed for when the time comes that the remaining 10 million dial up users finally cancel. The drop rate per year is well under what we thought it would be, though.

[SineSwiper]

It really depends on the size of the company and the field. Cable companies are known to be "agile" companies, where as companies like AT&T are slow as hell. However, that agility comes with the price of priorities changing every Tuesday. It gets really fucking annoying.

[Kupek]

Rewriting old systems is a risk - a huge risk, in fact. It requires significant effort to figure out all of the functionality of the old system, determine the structure of the new system, make sure it interopts with existing infrastructure, convert all of the old data and retrain all of the users. You don't know how much this will cost, or when it will be done. And I don't think this can be overemphasized: debugged code that already handles all of the edge cases is extremely valuable.

Compare that to something that is known to work (and where it doesn't work, workarounds are known), with known maintenance costs.

It might be worthwhile to take the risk, but it's not a given.