What on Earth is a Mainframe?: A review

I’ve been reading David Stephens’ self-published What on Earth is a Mainframe, (also available on Amazon) which is as close to z/OS For Dummies as we’ll ever see.

I deal with mainframes at work from time to time. I interacted with an old IBM mainframe of some sort when I was in college, using it to get on the Internet, do e-mail for classes, and write programs in Pascal. That mainframe has been gone almost 20 years now, but it’s more mainframe experience than most of the people in my department have.

That’s the thing. Mainframes have been on their way out for 20 years–which was why Mizzou retired Mizzou1–but they aren’t any closer to the door now than they were when I was in college. I wouldn’t call it a growth industry, but there are some tasks that haven’t managed to migrate down to smaller iron yet, and if they haven’t by now, maybe they never will. But the universities aren’t producing new mainframe administrators–ahem, IBM calls them system programmers–so while it’s not a growth area from a numbers perspective, it’s a marketable skill that isn’t going away.

That’s where this book helps.

Read more

What will you do with your Freedom Box?

Columbia Law Professor Eben Moglen has a great idea. He wants every household to have a server on the Internet.

It’s not as outrageous as you might think. The hardware exists today. and I’ve talked about it before. The Pogoplug is based on it. Right now it costs about $100. The trouble is making the software easy enough that anyone can plug it in and use it. But that can happen too, and that’s the part he wants to get done.

I can’t wait.
Read more

The 15-second rule and other (non) myths

The 15-second rule and other (non) myths

Cnet investigated some computer wise tales, myths, conventional wisdom, or whatever else you want to call it. The one I take the most issue with is the 15-second rule. They asked Geek Squad, and, as a long, long-ago Best Buy employee, the answer they gave to the 15-second rule is, well, what I would expect. Read more

What burn-in is and why you should do it

A friend bought a new computer this month. It arrived sooner than expected, and she called to ask what she should do. I said, among other things, to set it up and leave it powered on continuously for at least 24 hours.

A coworker who had never heard that advice asked why.

The reason is twofold. If a machine–electronic or otherwise–is going to fail, it’s more likely to do so very early in life. So running it for 24 hours will expose any parts that are due for premature failure. If the device survives its first 24 hours, odds are it will run for years afterward. But the second reason is called oxide healing. I can’t find a good link outside of electrical engineering textbook excerpts on Google Books, but basically, if any of the tiny electrical traces in computer chips (made of one exotic metal oxide or another, hence the name) are questionable, sometimes running continuously for a length of time conditions these traces and makes a borderline chip OK.

I do a burn-in on every computer I build. I’ve been doing burn-ins on my new computers for 20 years, and I rarely have hardware problems. I average about one hardware failure every five years, and that’s spread out among multiple computers. I do a burn-in on any other electronic device I buy as well, for the same reason.

Ideally, each chip should get a burn-in at the factory, and the complete device should receive a burn-in after final assembly, but that doesn’t always happen. Doing a burn-in yourself gives yourself some extra insurance, at minimal cost and effort.

It’s certainly not a bad idea to stress test your hardware, and frankly, I’d add a regimen of SpinRite to that list to make sure the hard drive starts its life out on the best foot possible. But if all you have the time and effort to do is power the PC up and walk away for 24 hours, that will give you most of the benefit.

Advice for setting up a computer lab

Two weekends ago, I headed back up to Bethlehem Lutheran Church to rebuild the computer lab I had set up for them a few years ago. Built on P3-based Compaq Deskpros and Windows 98, it had held up, but was desperately in need of repair.

The vision of this project was to set up a lab in a declining neighborhood, where kids could come to do their homework and adults could come to learn computer skills. The lab is run under adult supervision, and outsiders with teaching and/or training experience come in occasionally to teach computer literacy classes.

Ultimately, I want this project to produce some sysadmins in North St. Louis. Years ago, when this was in the planning stage, I told Pastor John Schmidtke that I wanted to see some Mercedeses and BMWs in that neighborhood, purchased with salaries earned from skills picked up in that lab.

Well, that hasn’t happened yet. But there’s still time. In the meantime, how about if I talk about what I’ve learned from three years of operation?Fancy computers not necessary

These were midrange machines when they were new. The lab cost some $10,000 to build, including materials to renovate the room. A contractor who attends my church renovated the room to my specifications. For what it’s worth, I told him there would be about 10 computers there, and the keyboards needed to sit so that a five-foot person’s elbows would make a right angle while seated. He built an around-the-room desk to hold the computers and bored out a couple of holes at each station for the keyboards.

Back to the computers. Today those computers would be worth about $100 apiece. That’s OK–they’re fine for word processing and e-mail, casual web browsing, and educational games, which was the goal. They were fine then and they’re fine now.

Any business with a philanthropic mindset could duplicate what we did here with a dozen outmoded office computers. A Pentium II with 64 MB of RAM can run Windows 95 or 98 adequately. There’s no point in messing with anything lower than that, since P2s sell for 50 bucks a pop these days. One thing I will say: The quality of the keyboards, mice, and monitors is much more important than the computers themselves. If people are going to be typing, they need decent monitors, a good mouse, and a keyboard that doesn’t feel like oatmeal. So if you’re working with a budget, plan on spending the most on monitors and optical mice.

Adequate power outlets completely necessary

There was only one hardware failure in this lab’s three years. One computer wouldn’t run, no matter what I did. It wouldn’t even power up.

Well, speaking as someone with a decade of professional experience in sales and service of computers, I’m embarrassed to say the computer that failed wasn’t plugged in. It looked like it was plugged in. But with the electrical outlets in a mismash and the cables all over the place, it wasn’t. Why didn’t I try another power cable and discover the problem? I have no excuse.

I suspect someone needed an outlet for something–be it speakers or a hair dryer, you never know–unplugged something at random, then didn’t plug it back in when finished. Worse yet, the business end of the cable ended up in a box under the desk, so everything looked fine. The result was a computer that didn’t run for about two years. I didn’t even find the stray end of the cable until I traced several other cables.

So provide as many outlets as your breaker box can handle. Provide extras. People will bring in other things they want to plug in. Three outlets per seat (computer, monitor, and speakers) is inadequate. To be safe, plan on four.

Create system images

You’ll have to reload the system no matter what you do. Lockdown software can be counted on to break things for the innocent and at best only slow down the people who are going to circumvent it. When the goal is for people to learn how things work, they need to be at least somewhat free to experiment. So make images of a working system with all software installed. Train someone to re-image the system. Then don’t be shy about re-imaging. Do it once a week or any time something goes wrong with a computer. If you can’t afford Norton Ghost or PowerQuest DriveImage, a freeware alternative called Savepart exists. Use it. Come to think of it, use that, and spend the money you’d spend on Ghost or DriveImage on better keyboards or mice.

Use identical hardware

System images don’t save you much trouble if your hardware is all different. Your image probably won’t work. Use systems that are as similar as possible: same motherboard, same video card, same sound card (if not on the motherboard), same network card. Put the cards in the same slots on each machine.

Frequently you can get away with not being that careful, but trust me, you don’t want the one time in a thousand that it matters to come and bite you.

Match your hardware and operating system

Sometimes it’s not possible to use completely identical hardware. In those instances, make sure any dissimilar hardware is recognized by the OS without loading any drivers. For example, this lab has a mix of Netgear and Intel network cards. The Netgear isn’t recognized by Windows but the Intel is. So I made the image for a Netgear, and when the system comes up with an Intel, it handles the situation gracefully.

Do the same thing for the sound cards and video cards.

Don’t ask too much of your systems

Yes, Windows XP will install and boot on a Pentium II with 64 MB of RAM. But there is absolutely no benefit to it. XP wants lots of memory and CPU power, and when it doesn’t get it, it’s a slow pig. Realistically, a P2 with 64 megs ought to be running Windows 98. Don’t even try 2000 or XP unless you have P3s faster than 500 MHz with 256 MB of RAM.

While there are subtle differences between versions, for the most part Windows is Windows. If someone can use Windows 98 and Office 97, it’s easy to adjust to the current versions. Don’t be afraid of running back-level versions.

Get discounted charity/educational software

You don’t have to pay $200 for Windows and $500 for Office to build a computer lab for a charity. I don’t know what charity or educational prices are right now, but it’s a fraction of that. The rule of thumb is this: If you don’t pay sales tax, you don’t have to pay full price for software either.

Most computer stores can get you information on charity pricing. Support your local computer store. You might need it someday. Who knows–the owner may be able to get you a line on some things you need.

And if worse comes to worse, run Linux. That was my original plan, but Pastor Schmidtke wanted to use what local businesses were using. That’s a better plan, but if you can’t afford software, Linux works nicely and is more than adequate for things like word processing and web browsing.

Just do it. That’s about all else I can say. The lab in your head isn’t doing anyone a bit of good. If it’s worth doing, it’s worth doing badly. If you’ve got a vision for this, get what you can to get the lab up and running. You can always come back and add newer or more computers later.

WordPress Appliance - Powered by TurnKey Linux