Slashdot Mirror
Burying a Mainframe In Style
coondoggie writes "Some users have gone to great lengths to dispose of their mainframes but few have gone this far. On November 21, 2007, the University of Manitoba said goodbye to its beloved mainframe computer by holding a New Orleans-style jazz funeral for its 47-year-old IBM 650, Betelgeuse. In case you were wondering what an IBM 650's specifications were, according to this Columbia University site, the 650's CPU was 5ft by 3ft by 6ft and weighed 1,966 lbs, and rented for $3200 per month. The power unit was 5x3x6 and weighed 2,972 pounds. The card reader/punch weighed 1,295 pounds and rented for $550/month. The memory was a rotating magnetic drum with 2000-word capacity (10 digits and sign) and random access time of 2.496 ms. For an additional $1,500/month you could add magnetic core memory of 60 words with access time of .096ms. Big Blue sold some 2,000 of the mainframes, making it one of the first successfully mass-produced computers."
to the guy(or girls and guys) who did this. Any machine that has been in service or at least functional for 47 years, deserves this kind of respect and this kind of send off.
Yes, i know it's only a machine, and it has no feelings. But this is a respectful send off, and 'job well done, thank you' to all people who were involved in designing, maintaining and producing this mainframe.
Plus...it's a very cool..and sounds like fun.
So the weight and exorbitant (extortionate?) cost of the CPU are known. So how fast was it, the CPU? And how come it continued to be used this century?
the 650's CPU was 5ft by 3ft by 6ft and weighed 1,966 lbs, and rented for $3200 per month. The power unit was 5x3x6 and weighed 2,972 pounds.
...And if they had recycled the copper and aluminum in just one of each rather than burying them, they could have bought an entire
lab of mid-range PCs with it.
But hey, that wouldn't get kitchy national media attention.
In fact a Millenium 1015 is quite a recent mainframe - introduced in 2000, (hence the name) although the 1015 is the bottom of the range unit with just a single processor.
It would be nice if reporters actually researched this story instead of merely cat'n'pasting the whimsical and completely inaccurate press release.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
I don't know anything for sure, nobody does. However it IS very impractical and redundant to start every sentence with 'As far as I know' or with 'I might be wrong here'.
That's understood, everything I type and know is relative to the information I have and the way I perceive it. For all we know you could be a figment of my imagination, or I could be the figment of yours, or we could be a figments of someone else's. While metaphysics are fun and seemingly profound and deep, they are ultimately pointless. The fact (question arises, what is fact and what is mass hallucination ) that computer cannot effectively communicate its feelings to me makes 'it only a thing that has no feeling to me'. Reality is not something stable, it's basically the general consensus of a large group of people. For some people God is reality, for others aliens.
So yes everything you said could be right, it could be just nonsense as well.
These arguments/discussion are pointless. The real question arises is 'Is it worth discussing?'
You said:
Do you really know that you are anything different than a little sim in a simulated world, or a self-aware mathematical entity in a mathematical universe? I could ask you 'Do you really know we are sims in a simulated world? What proof do you have?'And that would make this just a modified 'Does god exist argument.'...which are pointless as well.
So here are my final statements:
Given the current evidence, I can only conclude this is the only reality there is as currently there is no evidence of any other
Given the current evidence, I can only conclude computers have no feelings, as currently there is no evidence of computers displaying self-conscious actions
The truth is not absolute, no one said it was and who ever think it is, is a bit of a fool (and we are all fools in one way or another). The truth is just a group of conclusion we can make, based on current evidence. Everything I say I know is relative to the today's truth not the absolute truth...we will never be know the absolute truth...or at least we will never be sure we know the absolute truth.
A helping of desert is a lot of sand.
Care for some dessert instead?
My first Journal Entry ever, in 8 years! http://slashdot.org/journal/365947/aphelion-scifi-fantasy-horror-poetry-webzine
Do you really need the latest generation of hardware to serve web pages?
No. But you might find it is more economic to do so. If you can consolidate 4U of servers into 1U (for example), then it may be cheaper to do so rather than continue to rent the 4U of space (and it'll save power and generate less heat too).
http://blog.nexusuk.org
Hey, any excuse to party and burn something works for me!
More likely, 25 servers with one down = most of them broken, because the one that failed was providing DNS or external network connectivity or NFS serving or Kerberos authentication or the database or...
You can't assume that just splitting services across different machines will make them more reliable. Most of the time it makes them less reliable, because instead of a single point of failure you now have several points of failure and if any one of them goes wrong then your systems break. A mainframe is a single point of failure and if it dies, everything dies... but it doesn't tend to die because the hardware and software are designed to be more resilient than standard PC hardware and operating systems.
Yes, you can use commodity hardware and software and distribute your computing tasks to get good reliability, as Google does with its hordes of cheap servers answering search queries. But you have to be clever to do it. Just taking one system and splitting it into twenty-five interdependent systems does not add reliability.
-- Ed Avis ed@membled.com
I can remember sitting in on an IT meeting at a place where I was contracting (doing Netware Support) where one guy had to report back on his efforts to sell an old IBM Mainframe System that spanned the entire length of the computer room. The system had been replaced by this tiny, shiny, black AS400 that sat in the corner.
"Best so far is about £2000" said the man.
"You can only get £2000 for all that equipment!?" said the astonished IT Director.
"No", came the reply, "That's the cheapest to pay someone to strip it out and take it away!"
AT&ROFLMAO
That being said, I think the debate in servers vs. mainframes is long since over -- servers won, for the most part, except in mission critical applications where 100% uptime is mandatory. Servers are cheaper and with clustering you get extremely high availability and/or computation power to spare.
My blog
Mainframes are not primarily about calculation bandwidth (that's a supercomputer); they're about I/O bandwidth. If you watch the processor usage on your personal computer when it is slowed to a crawl, you'll see that most of the time the CPU utilization is not particularly high. That means the CPU is starved for I/O. That's why the lowest range IBM mainframe CPU, although not much more powerful than an Intel Core 2 Extreme, handles dozens of times the load except on compute bound tasks like cryptography (which is handled by special coprocessors on the mainframe).
You can build a supercomputer by clustering relatively weak processors, however that supercomputer is limited to problems that can be efficiently parallelized. Fortunately that set of problems is highly useful.
Similarly you can attempt to build a mainframe out of low end boxes hooked up to high bandwidth storage, but you have similar limitations in the I/O domain: you are hooking up a bundle of straws to a fire hose. If individual processes need more than one straw's worth of I/O bandwidth, this approach does not work.
Arguably you can most often find a way to break down a task into small I/O chunks; some tasks like Internet indexing and search fall naturally into that paradigm. Other tasks that are transactional in nature and require certain global semantic constraints to be enforced can be distributed, but with considerable overhead and complexity in application design and system administration.
The best reason to go with a cluster of cheap servers is incrementalism. If you're up to your eyeballs in servers you can just reorganize them as you suggest, without getting all new software and IT staff. If I set out to create an application that takes twenty-five to fifty individual servers, I'd definitely look at a mainframe as a host.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
My first thought was that if we personalized computers more, perhaps we wouldn't waste as many of them. We have become very much a disposable society, in which the strangest part of this is that anyone bats an eye about the loss of a computer. Yet I remember when we used to mourn the passing of many of them. A lot of our waste problem in the world is caused by our willingness to assume that disposing of something does not require ceremony and can be done as casually as exhaling a breath of air... except no one is recycling the air and it's getting a little stuffy in here.
It's one reason people have big weddings... to make it so expensive that you think twice before throwing it away on a mere argument. If throwing away a machine were more expensive, maybe we'd think twice about doing it... or better still, about buying one in the first place.
Yes, it would hold back progess. But where is progress leading us right now? With luck, we'll have computers powerful enough to solve the problems we created by having computers. And without luck, we may poison our world and all die. Ah, yes, the smell of progress is all around us.
Kent M Pitman
Philosopher, Technologist, Writer
I am the proud developer of the actual site that TFA is linking to, you can see it at: http://umanitoba.ca/mainframe We are all amazed by the popularity of this event! And just for the record, the mainframe was RECYCLED so don't worry, we are very environmentally friendly here at the U of M!
The metric system is not any more "logical" than the imperial system or any other. There is no "logic" in a meter, an inch or a stonetoss. The whole point is that it *is* more practical.
Yeah. And in imperial, everyone is 5 to 6'6". Big deal. I don't see how ft/in is simpler here.
You show me the cup which is exactly
As opposed to 1 l (which happens to be 0.1m^3) of water which has a mass of 1 kg. And to accelerate it by 1 m/s^2 you need 1 Newton of force. Which takes 1 Joule when you do it along 1 m. I'd love to see how you calculate the force it takes to constantly accelerate a pint of water to 1 mph over a distance of 1 ft.
While there is some truth to that, I'd still like to point out that it is overrated. What matters is the how many different prime factors a numeric base has; in case of 10 we have two (2 and 5); in case of 12 we 3 (2, 3 and 4); and in case of 16, we only have one (2). A base-10 system does not work better with "computers and exponents" better than any other. There are people saying that base 12 would be better for general use than base 10, but I believe that the difference is not that big after all.
What matters *more* though is that pretty much *anything* else uses base 10, and thus the choice of *any* base except 10 is a bad one, because it makes interaction with with those systems more difficult and because people are used to base 10. What's half of 3 hours, 7 minutes and 24 seconds? What's a third of 7'6"?
The Imperial system is worse in all aspects which matter. End.