Slashdot Mirror
IBM Wants CPU Time To Be A Metered Utility
kwertii writes "IBM CEO Samuel J. Palmisano announced a sweeping new business strategy yesterday, pledging $10,000,000,000 towards redefining computing as a metered utility. Corporate customers would have on-demand access to supercomputer-level resources, and would pay only for time actually used. The $10 billion is slated for acquisitions and research to put the supporting infrastructure in place. Will this model revolutionize the way companies compute, or is this plan doomed to be another PCjr?"
Ballpoint pens proclaimed "the wave of the future".
What we call folk wisdom is often no more than a kind of expedient stupidity.-Edward Abbey
It will be tought getting quarters and dimes in the floppy slot. Or is that a cupholder?
GetTheJob.com : Nothing but Real Jobs.
This won't revolutionize anything... I remember this when it was called timesharing on mainframes. The revolution was moving away from that model...
"Timesharing" back in the early days of computing. I always assumed too that this was the bank's way of justifying the abominable practice of charging ATM fees.
Why do I M2 everything negatively?
It will be a revolution until Linux becomes mainstream on the desktop and every computer on the corporate LAN is part of a cluster, when users log off the computer re-joins the cluster. Companies should look at what they already have before shelling out more money.
As long as IBM doesn't change the rates for processing in conjunction with Moore's law, making the processing cheaper on their end, there could be quite a boost in profits from their current business model simply selling the hardware.
But then again, one of the reasons that Enron went down is that they quit selling real, hard, physical commodities and instead went directly to a more ethereal model of paper sales and transactions.
So will this move the ball towards corporate responsibility in this area?
I am certain that a lot of companies will try to avoid it if at all possible. Of course, this would be controversial, especially re: open source, etc. but it is not the most common practice now.
"It is a greater offense to steal men's labor, than their clothes"
In the mainframe world cpu cycles are already a potential billable transaction..
So the concept is old and crusty..
---- Booth was a patriot ----
Is this a hint that the industry has finaly figured out how to get the bugs out of software?
Samuel J. Palmisano announced a sweeping new business strategy yesterday, pledging $10,000,000,000
I think Samuel has been watching Austin Powers way too much.
Maybe you should educate the morons of tomorrow so they'll stop believing the leaders of tomorrow. - Dogbert
If the huge false start that was Application Service Providers showed anything it is that corporate customers don't trust computing resources that are outside their control. It doesn't matter if IBM can provide a better service or a more reliable one, it just doesn't feel that way - and the IT guys will never report favorably on something that will put them out of a job.
It's PCjr, it's Gavilan, it's all kind of failures. And $10,000,000,000!
for companies and institutions that use a lot of heavy computation.
It takes a lot of time, space and know-how to own and maintain big-@ss computers. With broadband connections being commonplace, you could run your own progam remotely, and let a specialist (like IBM) handle all that stuff. And of course, there is value unlocked by having multiple users share common resources.
Of course, the vast majority of companies and institutions (not to mention individuals) use their machines mostly for word processing and surfing the net - and thus they will have little use for this kind of service.
Tor
More like "charge you to get back your bits. The ones in your computer and your pants."
Soko
"Depression is merely anger without enthusiasm." - Anonymous
...in an age were processors are dirt cheap anymore. I mean really, if I saw a p2 400 chip and a quarter lying side-by-side on a street corner, I'd pick up the quarter.
To win back that initial 10 BILLION [pinky finger to lips] investment? $1 per Ghz?
$.01 per Ghz?
"Draco dormiens nunquam titillandus."
It took the rest of the computing world YEARS to match the color & sound of that baby. What, you don't remember CGA and speaker music? Tisk.
This reminds me of how scientific super computers/clusters are leased out to various researchers for computing time. It works very well for that target audience, or so I am told. With a well formulated design and clever marketing IBM could find another audience for such "metered computing". I can't think of any other industry outside of the scientific community that would be looking for such a solution though. In the long run, wouldn't it be cheaper just to hire a consulting firm to build you a cluster?
...but, as the article points out, only for corporate/supercomputer types of situations.
After all, the PC revolution demonstrated that individual users want unrestricted computer usage on their own terms, and were/are willing to pay a fairly generous amount for it.
I only see this project working out as long as companies see it as cheaper than building their own solutions. Linux-based clusters can provide a fairly low-cost solution for a lot of high-end computing needs (like rendering tanks) -- that's what IBM will be competing with.
I think it'll boil down to how greedy IBM gets on pricing. If it's too pricey, companies with a fairly regular need for lots of computing power will deploy their own internal computing clusters -- which is ironic, considering that IBM will probably stay very interested in supplying such solutions. It sounds like they're just trying to play all sides of the game: Sell the big/pricey hardware, sell time on the big pricey hardware, sell the lower-cost alternatives -- or at least the contract to deploy and maintain them.
Xentax
You shouldn't verb words.
"Will this model revolutionize the way companies compute, or is this plan doomed to be another PCjr?"
yes. Doomed to be another PCjr. People want expensive goods that they can brag about. Plus, let's see you game on it. Personally, metered utilities are bad enough on their own without extending into my computer.
In SOVIET RUSSIA... erm...NSA AMERICA, the Internet logs onto YOU!
Actually people are already doing this. I work for a Fortune 500 company that outsourced all of its data centers to IBM already. IBM charges them based on CPU and disk space available for use at any time. This will allow them to cut costs even more by only paying for what they actually use; there is no expense for idle-time. Yes this is a concept from the 60's but everything gets recylced. Another concept from the 60's is a fat server with dumb terminals. In the 80's we went to a PC on every desk and now because of networks and the internet we have gone back to a fat server and dumb terminals.
Looking for a job?
Want your resume written professionally?
DON'T USE TUNAREZ!!!
See, at that time ubiquitous networking was not a way of life. Also, software engineering was not as mature as it is now WRT to virtual machines, encapsulation, OO design, etc.
Of course, all those technologies did exist then, but they can be counted on to be everywhere now. The reason mainframe timesharing gave way to PC's is because PC's could provide a more flexible and convenient sandbox to compute in, rather than the cumbersome interface of working with the mainframe in the company basement.
These days returning to the idea of computing power as a fluid resource is a good idea because the landscape has changed and the world might actually be better prepared to accept the tradeoffs since the tradeoffs are much less significant now.
don't ever let this happen. The car design scenario creeped me out. I work for an automotive supplier, and we ALREADY have to wait in line to use test equipment, testing chambers, etc. I can only imagine the local supercomputing hub monopolizing the speedy machine, creating more lines to wait in, and IBM bringing its supercomputer prices out of reach for anyone but their own subsidiaries to purchase. Could be a disaster, indeed.
hi, I like pancakes -.-- -.-- --..
The concept IBM is going for is to treat IT as another utility. Instead of some small company having to keep an expensive IT staff and maintain their own computers/network/storage, IBM says that it will do this for you. IBM will essentially replace the IT department and let some organization concentrate on running their own business.
The cost saving of such a model (if successful) are quite substancial and will save everyone money in the end.
I think IBM is on the right track with this and they are the only company really positioned to do so.
If this isn't an idea for a killer app, I don't know what is.
With IBM's continuing support of Linux in the commercial and high end server space, I have no doubt that this will be a GNU/Linux friendly project, if not composed entirely of GNU/Linux software.
And just imagine the possibilites for breaking the MS monopoly. I can just imagine companies with hundreds of cheap, dumb, never-needing-to-be-upgraded X terminals connected to this computing "utility" for all their office/CAD/research/calculation/accounting/etc needs.
Why not combine your computing "utility" bill with your software "utility" bill? IBM's supercomputers could always have the latest versions of Sun/Open/IBM/etc office suites. It would be the natural extension of the software subscription model.
This project is going to make MS quake in its boots .
Nowadays the need for computing are everywhere. For instance where I'm studying almost every research group has its own Linux cluster. The use of these hardware comes in burst--often it sits idle, but when it is needed, it turns out to be under-powered. Besides, the maintenance of this machine costs too much man-power and money. After a couple of years/projects, these machines go out-of-date anyhow. Of course any given lab could always negotiate the use of supercomputers at large research centers, but IMHO having the infrastructure in place will be very useful.
Likewise you can get a machine with an big ol batch of CPUs, most of them disabled. Over, say, the Christmas rush you call your salesperson and have the other CPUs turned on for a month. Again: Strange but the corporate customers seem to like it.
I doubt it'd affect Joe Average Desktop user all that much. Your average desktop has more processing power than he'll ever need and is already dirt cheap. It's only when you start talking machines worth millions of dollars that this sort of thing makes sense. The same people who go for this stuff pay out tens of thousands of dollars a month for support costs and they get some very good value for their money.
Many of you youngsters might not be all that hip to mainframe culture or mentality, but it's a pretty good deal and those machines are still amazingly fast. A lot of shops haven't been able to get rid of their big iron because PC clusters just couldn't deliver as promised. Our VM box back at school routinely had 5000 users on 13 years ago and that machine never even hiccupped.
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
The last 35 years development in computers were precisely to move away from the "metered service" model which made IBM's fortune.
On will recall that IBM's data-processing customers since the 1920's were charged by units of information stored/processed by the way of forcing customers to buy Hollerith (punch) cards solely from IBM, and run them in rented machines whose rental price was directly proportionnal to the throughtput of those (a card reader that processed 600 cards per minutes cost twice as much as one that processed 300, yet the only difference was the size of the pulley off the main motor - and you could upgrade by having an IBM tech that came and changed the pulley for a bigger one).
So is it that the ghost of Thomas Watson sr has made a comeback to IBM's board of directors????
We had metered cpu usage at college. It was a constant, annoying nightmare. Though the "money" was supposedly "fake" to students, you had to beg the admin assistants in the CS department to get more when your account ran low. The administrators of the Computer Center claimed it was actual money charged to each department. The school also gave out free accounts to students with small money allocations in them which gamers borrowed and stole to play GalTrader on the VAX.
I thought it all went away until I started working for IBM. Every time you log out of the mainframe the computer told you how much money your session cost the company. That turned out to be real money that was charged to the department you worked for. We eventually reverted to using X Terminals connected to massive, rack-sized RS/6000 machines instead of the mainframes after that.
Kris
Kriston
If my history books and gathered information is correct, that was a business model used in early computers. A company would lease CPU time to users, generally because the end users couldn't afford the massive costs involved to purchasing and maintaining them.
Now, I'm relatively young (mid 20's), but I recall people not even a half generation older than I telling stories about getting in trouble for running up large bills on their school's timeshare account.
I could see where this might be useful, but only for a small handful of customers. There are not very many users of supercomputer's out there right now. I can't see that number increasing much just by servicing new customers who could benefit from a supercomputer but couldn't otherwise justify it for a short term project.
If they are dumping 10 billion dollars into this, they must think they are going to get at least that much out of it. I seriously doubt that they could do so, not without ridicously overpricing their service. For small time users who don't need supercomputer levels, there are much cheaper ways to go. (Buy your own gear, lease your gear, etc)
I work for a specialized outsourcing outfit that manages storage for large customers (internet datacenters primarily). I know how much of a pain in the ass it is to accomplish what we do now. I could just see the mess people would get into by getting into a timeshare system like this.
A slip of the foot you may soon recover, but a slip of the tongue you may never get over. -Benjamin Franklin
Computing will be made into a utility like hydro and the closest most users will get to a computer as we know it will be a wall socket. This will happen because big business and government want it to happen, because like bank robbers to banks, they know that's where the money is. The only recourse will be to go off the grid. Like many farmers are now going off the hydro grid and turning to wind and solar power we will have to go off the grid perhaps forming co-ops, credit unions and other institutions to allow us to access the big business and government run net while not being subject to the strictures. Maybe an independent satellite grid?
"Academicians are more likely to share each other's toothbrush than each other's nomenclature."
Cohen
One of my first computers was an IBM PC JX (which I understand is similar to the PC JR). This was the first Personal Computer (ie. box with drives, monitor, keyboard) that I ever encountered - a big step up from the old cpu-under-keyboard micros. The 256K RAM was a big step up too.
It was made in 1980 AFAIK, but had a 3.5" diskdrive and a cordless keyboard (features which never came along in other computers for several years).
Why does the article talk about it as if it was a bad thing?
So how will corporate customers be convinced to trust sending their data to and processing their data on IBM's grid of computers?
But it'd be nice for running multiplayer gaming servers.
Is that we continue to see companies like Microsoft and IBM looking to change their revenue model to subscription based services. It makes sense, just today I was talking to a friend about what parts I was planning to order to build his computer. And thinking about it, the average user can run most of their average software on a 1ghz intel or athlon board. Microsoft is having a problem getting people to continue upgrading simply because the lifespan of the software as-is is good enough for most. Naturally, the hardware demand will slow when software isn't written in such a way that it requires more horsepower. I think these companies see the writing on the wall. I'm just disappointed that instead of Revolutionizing they would rather rope consumers into some sort of model that doesn't require any extraordinary efforts on their part.
This reminds me of UNIX's parent, Multics, which had similar goals but never achieved widespread acceptance.
A CRAY supercomputer in 1980 has the equivilent processing power of a 500MHz processor. By the time IBM gets people to switch to this "pay for cycles" method computers will surpass it's ability in the cost/performance arena.
It's not always how fast you go, it's how efficiently you get there. I could fly on the shuttle from Kennedy Space Center to Edwards Space Center (assuming NASA would lighten up the travel for free restrictions on italians in oklahoma! lol) but as fast as the shuttle goes I could drive there faster (although not nearly as stylish).
Now I just need to get a solar array to power my array of older computers so I can sell back their CPU cycles to IBM and maybe, just maybe, earn enough to pay for the solar cells.
Kevin Fox
No, this is _way_ more expensive than the PCJr was. $10billion? Sheesh... Lot of money. I wonder how they came to that figure? Why not 9 billion, or 11 billion?
At least the PCJr wasn't doomed to begin with -- the only way to make CPU time valuable is to limit the amount available. With Moores law and economies of scale (how long till we have an 8-way 5GHz CPU system? How much longer until we have the same with 10GHz?) I find it difficult to conceive of any way to beat it, other than absolute domination.
"But actually trying to use m4 as a general-purpose langage would be deeply perverse" --ESR
PCjr
Everyone's so incredibly focused on their quarterly/monthly/dayly/hourly earnings numbers that companies don't want to sink big bucks into big IT project right now. Give them a chance to rent something by the hour/MIP/whatever and even if they pay more over the long run it keeps the 'up front' expenditure down and doesn't hit this quarter's numbers...makes the numbers look better, keeps the investors happy, lets the execs sleep a little better at night. Plus if you rent MIPS for a month on some new super-duper project and realize it's a dud you can walk away without having invested too heavily.
I don't know what everyone is comlaining about. Take a look at the wonderful user-oriented, monopolistic services companies like the Phone and Cable companies currently provide (Qwest, Verizon, Cox, etc). This is a terrific model to emulate. And think of all the lovely intelligible Taxes the government could add to your monthly computing bill.
0xfeedface
If all it is is code for a price hike, yes, it will fail.
Want to Know How to Cheat the GPL? Read On!
You save money by outsourcing to IBM. But then you have other things to look at.
First you'll still need some sort of helpdesk staff. Internal or outsourced to IBM.
Second you're going to be spending more money on telecom circuits. Now you'll need enough bandwith out to the internet to support all of your "knowledge workers."
Third security. Who will own the data? How will the data be secured against competitors who might also be IBM's customers?
Fourth is backups. What is the liability if IBM can't restore a deleted file or email? What about redundancy and downtime? Who is responsible for lost revenue?
Fifth it won't save as much money as IBM is hyping. Every company has tons of data that is rarely used, but still sits on file servers taking up space. This model won't change this. You will still be paying for storage that rarely gets used.
wow, bad programming would _REALLY_ cost you!
there goes the wintendo TCO
EverlastingPhelps wrote:
n ni um/mgoals.html
;)
> set up a "trusted" system and then use your
> Passport account to charge you by the bit.
You mean like this "trusted" system?
http://research.microsoft.com/research/sn/Mille
Yep, Microsoft already has their Millenium planned out. Of course, Godzilla already has their destruction planned out.
"At this moment, it has control of systems all over the world.
And...we can't do a damn thing to stop it."
Miyasaka, "Godzilla 2000 Millennium" (Japanese version)
Godzilla's 48th Birthday will be this Sunday.
There are some good reasons for selling CPU time as an on-demand service. I'm sure IBM knows what those reasons are and will use them to try and sell this concept.
But there are two, possibly three, very powerful forces working against them here.
First, computing power is very cheap these days. It's not precious. People have 2 GHz Pentium 4 processors sitting around waiting for their next keystroke in Word and they don't feel guilty about wasting CPU cycles.
Second, the price keeps dropping at about a 40% annual rate. That same cheap PC waiting for the next keystroke would have been worth tens of millions of dollars to a scientific establishment in 1974. Not now. With a market where the supply of computing power is constantly increasing, it will be very difficult to peg any kind of price that people can use to make buying decisions, because those decisions will look foolish a year from now when someone asks why they didn't just buy a couple more PCs, or even a rack of PCs to do the task.
Third, the rented computing power needs to be connected very well with the data it will be processing or producing. If the rented machine is on the customer's site next to his SAN warehouse, then everything's fine and this may not be a real problem. But if the big machine is in Fishkill and the customer's 10 TB of data are sitting in a weird database inside a firewall connected via T1 to the Internet, then there may be a problem.
If I were IBM, I'd look into ways of increasing demand for computing power. Protein folding simulations for new pharmaceuticals is one way, financial scenario analyses is another, and database mining is yet another. They have to make customers want to buy extra computing power because they can easily see a business need for doing so.
The other thing is they need to increase demand for the ultra high reliability mainframes. For some of those computing needs, a rack o cheap PCs is going to be a much more economical choice for their customers. However, there are some applications, like VoIP telephony, video streaming, or credit card approvals, where people would get upset by downtime.
"Provided by the management for your protection."
The initiative is expected to cost $1.86e+93 Kabillion dollars.
If Jesus wants me it knows where to find me.
I think many people are missing the point. This is not a return to Mainframe-style time-sharing (although the technical descriptions and business model might seem that way).
What IBM is proposing is that companies should not have to deal with running an IT department, when all they want to run is their business. They can simply pay for CPU cycles just as they pay for electricity, and their applications will simply use those cycles to perform their desired computation/storage.
Think about this: No more dealing with hardware. No more huge IT staff. No more complex budgeting for IT. No more upgrade nightmares.
Also, companies with as weak IT department will now be confident that the IBM (or whoever) datacenter folks will handle all the security concerns for their application (user access, encryption, authentication, DoS, hackers, etc). Likewise, they will feel confident that the datacenter folks will mirror and backup their data offsite in the event of a catastrophe, something only large companies today can afford to do.
Once companies realize the benefits of this, not only will they rent CPU cycles, they might even decide to rent applications as well. Today the Applications Service Providers model has not taken off due to a lack to a killer app. I think Grid Computing is that killer app.
What IBM has said is that it hasn't got anything new to report but that its still here. If you look at their figures $10Bn works out at 3.5bm for the consultancy firm they purchased, a few billion for Grid computing, and I guess a couple of billion for linux. With a bit of spare change for research.
Why are they doing this? My guess is that CFO's keep complaining about the cost of computing resources. A multinational with 10,000 desktops still has to ask for clusters and supercomputers for serious work while TFlops of processing are sitting idle on the secretarys desktops. Hard Disks, which used to be able to just about hold the OS, Office suite and files now have 10's of GBs of wasted storage.
If you're serious about using computers you want to use resources efficently. And from IBM's perspective so how does this idea sound ...
IBM sells computers to a firm, it then sells the software to turn all their hard disks into a P2P file storage system so that you never lose that important document ever again. Instead of a new cluster - set all the desktops to process data overnight as a massivly distributed system. (using IBM software), installed by IBM engineers under the direction of their new consultants. And of course the only real option for this is Linux.
A single, nice, neat package. A single point of contact and massive economies of scale. Now assume that their contract allows them to use/sell spare cycles and their revenue stream suddenly improves a lot.
One online magazine did all that? Now I know who to blame!
In any case, I'm not sure how far this return-to-the-mainframe idea will take us; we've had the technological framework for doing this for years -- think RPC, OpenStep's Distributed Objects, Sun's GRID engine -- but where's the real value to the department's bottom line?
I spent a number of years working on an extremely computationally-intense business process for the not-so-late, not-so-lamented WorldCom. For about half of that time, I was running the systems architecture and administration group, so performance management was a huge concern. We chewed up a lot of user time, but we were primarily hampered at every layer of the process by I/O (disk and network) and memory constraints. The same has been true of the accounting and provisioning systems I've worked with since then: the enterprise-level bottlenecks these days are things that can't be purchased on demand.
I'm sure there's a market for these kinds of services -- medical imaging, for example, though the network costs would be high -- but something to bet the Big Blue (computing) Farm on? I just don't see it. *shrug*
"Freedom is kind of a hobby with me, and I have disposable income that I'll spend to find out how to get people more."
A lot of people here are pooh-poohing this as "time-share" computing which was around back in the day saying we've moved away from that concept. I think it could certainly be a viable option for companies that are wanting more computing power, but also looking to cut costs.
Also, consider that the companies making use of this would never have to upgrade their own clusters. I constantly see newer clusters being planned by companies and governmental agencies. It's always more processors, more MHz per processor and more nodes per cluster. Why not offload all of this onto a company (IBM in this case) who can put the resources (both in hardware and personnel intimately familiar with that hardware) necessary to maintain and grow ever larger, more powerful clusters.
IMHO, it seems like a great idea. It will give far more companies access to "super-computers" than ever before and at a significant savings.
It seems that once again IBM is being a very forward-thinking company and will probably end up make a pile of cash because of a little foresight and some guts to act on it.
Co-founder and designer at Music Nearby: http://musicnearby.com
With the millions of PCs sitting around with "Idle" as the busiest process, imagine getting a penny or so per CPU second, selling CPU time on the open market to the
Now, you have a value for CPU time on the open market. You should be able to
For a reference of a future society that uses this, see Greg Egan's "Distress" and perhaps "Permutation City"
Design for Use, not Construction!
Lets get away from a world were nobody knows what to do with the computing power that is almost freely available. Let's move into the wonderful world of GRID COMPUTING, where we can con people into paying for something they really don't understand. GRID COMPUTING, the marketing way of saying that Beowulf is a book - now give me the money.
Also, and I believe this was addressed in the article, look at all those failed startups that were left with expensive hardware and koosh balls. With a way to rent time on a supercomputer, certain kinds of startups become a much more economical possibility.
By way of comparison, I once visited a company that does high-end CG for movies, videos, and commercials. They showed me their render-farm, which they said they supplemented by renting computers to lighten their load at busy times.
For a company like that one, IBM's plan would further eliminate the need to rent the space to stick those computers, the cost of running an overpowered AC system at less than full-power (when they don't have so many machines going,) the cost to order, ship, hook up, and maintain the machines.
If IBM's system were available and I were to start a CG house, I would get some smart programmers, animators, etc. and some modest workstations, but rent time as needed from IBM for creating finished product. (Just a toy example, if you're ripping my "business plan" apart, you've missed the point.)
Finally, many industries have a tendency towards consolidation. Larger and larger companies could be a good market for systems like this one, simply due to their size. Also, larger companies could be a good market for IBM because they make bigger moves when they enter new markets. (Look at IBM itself, committing $10 billion to this project.)
I love this idea. For AI issues that only need a burst of CPU time in a lot this would be a god send to say, counter-strike developers. They could write a really complex AI to auto-generate waypoints for bot behaviors but would only have to pop say, $50 for a thousand cpu hours worth of computing time. For game developers that want AI waypoint generation this could be a great tool to offload cpu calculation that only need to be done once in a while (i.e. Everquest adds a new, complex zone. They have a waypoint script that runs but normally takes 4 days to run. They pop a quick $200 bucks and kick out the results in an hour. Saving time and allowing better generation by being able to do massive complex tasks quickly.) Or what about a VERY complex random map generator that simulates thousands of variables down to what was for dinner? You can spend more time developing rich complex algorithms and you can choose to take 3 weeks to generate the map at home, but if your a comerical application (or a well funded server) you can have the the super computer spit a unique map out (i.e. for every X dollars dontated through paypal we'll generate a new custom map.) I can see it now.... drool...
-=[ Who Is John Galt? ]=-
If they came out with a toolbar for other browsers they could *really* have coverage.
...Is this the Future of PC Gaming?
Campers pay by the minute!
#2 has FINALLY been solved!
... *cough cough* Yes Bueller? Sell the usage of said supercomputers on a metered rate? Congratulations, A+ for you!
(1) Create supercomputers that get increasingly powerful and versatile, advanced to the point that they could defeat one out of dozens of chess champions.
(2)
(3) Profit!
Just because you can mod me down, doesn't mean you're right. Shoes for industry!
Either that or Carl Sagan!
okay.. who else needs a supercomputer? And don't say chess guys... IBM already does their computers.
Seriously.. where's the market for "part time" supercomputing? Who needs to simulate to the interaction of atoms on tuesday, but not on wednesday? I may be missing something.. so while i'm sounding snotty... i really am wondering who "part time supercomputer users" are. Can anyone enlighten me?
All the users i can think of also spend a bunch of their time hacking the setup of their systems as well.. the last thing they will want is to lose all that mindshare - because these guys are uber-deep into the understanding of the problems AND the MP architectures needed to support the specific problems.. OR they'll spend their time trying to explain their problems to the IBM guys.
Help me... who does IBM think the users of this power are going to be?
Okay.. the protein folding guys.... that's 1....
guns kill people like spoons make Rosie O'Donnell fat.
this would sure as hell put an end to distributed computing.
unless someone had the cash to cover for them. or they would be allowed for free (like an 800 number).
THERE IS NO DATA. THERE IS O
the dollar with her face on it was withdrawn and replaced with the gold one, * which doesn't have her face on it.*
Sheesh. Pay attention. Don't you even watch late night cable?
Can you say: Sack-ah-jew-ee-ah?
I knew you could.
They're very useful too, considering that a dollar is what a quarter was . . . several months ago.
KFG
KFG
Well, this is going to lead to a few possible outcomes:
1.) The first outcome is one in which IBM wastes $10 billion and becomes the laughingstock of the industry. This seems to be the most likely outcome because using Linux and Beowulf, anyone can assemble their own supercomputer for a small amount of money.
2.) People actually buy into this shit and start using IBM's model for processing. People obviously don't want to waste any processor operations -- similar to gas consumption in cars -- so there's an ongoing race to create efficient software (and it's about fucking time). Of course, this leads to a situation where as software becomes more and more efficient, it requires less and less processing power, meaning it can run on smaller systems, meaning IBM will have a bunch of supercomputers sitting around doing nothing because they've evolved themselves into obsolescence.
Did this CEO work for Microsoft at any point? This whole "strategy" (I'm reluctant to apply that word to an idea as bad as this) reeks of something that would come out of Redmond. Have they learned nothing during their embracing of Linux? Do they really think that the end user wants to pay on a per-usage fee? The power of the computer is that all I pay for is a connection and electricity...given a choice where I have that, or a system which also requires me to pay for processing time, and I know which one I'll go with, every time.
You can't simply add a cost like this to the cost-of-ownership of a product with no significant improvement in overall cost or performance and expect it to be widely adopted.
And here all this time, I thought it was Microsoft that played the "let's throw shit at the wall and see what sticks" game of product development.
Time to sell my IBM stock. They just jumped the shark.
blog |
Instead, you will have
"MY NETWORK IS BROKEN WORD WONT START"
every time the IT staff goes to make an upgrade on the server. That's the problem with having one site that houses all of the applications; when that site is down, everything is down. RAID drives are ok for individual redundancy, until there's a glitch and the motherboard melts (happened at my company), then you better hope you have backup hardware in the closet (which they didn't).
It only takes one misconfigured router in the right place to knock out a whole company. If IBM has their way, one operator error could take down a city.
Let's say IBM is able to set up a way to do what they propose, here's some basic utility concepts I'm curious how they will adddress:
1. Transferring product from generator (IBM supercomputer) to location. If you've just used 1 month of supercomputer time to model DNA folding, how will IBM transfer that data back to you? What if the computations and use are faster than the transmission rate? [Modem vs. DSL vs. T1 line]
2. Dependency - you rely up on natural gas and electricity to be there, and yes they go down, but can they guarentee their utility won't have worse problems - especially if its Windows run and goes down once a week, cutting into your bought utility time.
3. Regulation. Most utilities are regulated, and those that were deregulated have not always worked out for the consumer. Let's say company A gets rid of its expensive infrastructure for computing resources and uses IBM's utility. What if IBM becomes the only utility and charges way more than it should - there's no competition so Company A can't shop around. Along this same vien, if Company A is smart enough, they'll never enter into a utility agreement with IBM if they can generate their own computing cycles and be sure that they'll always be there, versus putting all their eggs in one basket.
IBM's idea may have merit, but anytime someone throws out the idea of a new Utility, that suggests that the resource they're selling is mainstream and essential, and therefore, is treated as a commodity. Those commodities are regulated and made reliable so that they never go down. I can't see supercomputing cycles as being something that is commodity, or for that matter, something I (or any company) needs to buy on a metered basis.
-When going for broke, go for Ithaca!
"The concepts of grid computing are not entirely new or unique to IBM. Hewlett-Packard Co. is pursuing similar ideas, for example." The original submission asked if this would "revolutionize" computing. I doubt it. This is one large company trying to react to another large company. For a company who outsources IT, this might be desirable. I don't think this would be good as a home computing model. However, as someone once noted "There's a sucker born every minute."
"Give a woman two glasses of wine and some pad thai, and they'll agree to just about anything." the Sports Guy
I agree; we won't see the revival of timesharing anytime soon.
The PC revolution was based on the desire to get replace dumb terminals with something that could do color graphics, fancy fonts, and WYSIWYG word processing. This evolved into a more user-friendly interface for data manipulation.
For data-intensive applications, timeshare computing was economical, and it worked over low speed connections. Back in the 80's, it didn't take much data to qualify as "data intensive", either. I seem to remember something about a 32MB hard disk limit, for those PC users lucky enough to have hard drives. In general, data was never shared with anyone unless a mainframe was involved. File servers eventually brought data sharing to the PC, but even then, record locking was a joke compared to mainframe capabilities. You could run quite a few dumb terminals over a 9600 bps line, but that is inadequate for even one web surfer today.
OK, what has changed? Is there some new generation of CPU-intensive applications that requires far more CPU power than desktop computers have? I think this is yet another case of a solution in search of a problem. The NetPC was supposed to run apps without the need for a hard disk. The concept died when people discovered that hard disks were cheap and broadband Internet was not living up to the advertising claims. Along the same lines, who needs supercomputer resources when none of our applications are really CPU-bound in the first place? Aside from specialized stuff like ray tracing, animation, and possibly busting DRM algorithms, I don't know how timesharing would become a mainstream product.
...instead of sitting around surfing the 'net on company time, people will sit around and do ray-tracing on company time?
You are in a maze of twisty little passages, all alike.
The problem, it seems to me, is that CPU's are so darn cheap now. Barring SETI@Home and such, when was the last time that the processor in your desktop/notebook/PDA needed more power, and would have benefitted from a connection to additional computing, or even supercomputing, power elsewhere?
Let's take it to an extreme. Suppose you have access to two computers, both with equivalent Internet access:
- A 1.3GHz machine with 128MB of RAM and a gigabit connection to your own private network of Cray 2K3's across town;
- A 2.6GHz machine with 512MB of RAM and no outside resources.
Which is going to do the kind of work you do faster? Seems to me that it will always be cheaper and more effective to add processing power to the machine that you have, rather than add connections to machines that you don't.Of course, this only applies to CPU power, not to information. If IBM applied these concepts to some sort of hyper-efficient all-inclusive datamining resource (think Snow Crash's Central Intelligence Corporation), I think they could make the Web as we know it look like a used comic-book stand. Just give me a Quad-Clawhammer PC with 4GB of RAM, a TB of RAID, and a VDSL connection to IBM Global DataMine, and I'll gladly take out a second mortgage.
I've always believed that most of the software run on the big iron was problem-specific and machine specific. IBM must be planning to provide some custom programming to make this fly, and I'm guessing that's an important part of the profit plan.
See what I've been reading.
What is it about this industry that people have to reinvent old ideas that did not work when they were brought out originally? The concept of paying for what you use is not a new idea.
Heck, it is HOW THE INDUSTRY BEGAN. Colleges used to pay to use the computer, and you had to schedule time and code by entering in the values and then execute the program. If you were lucky, you could use this nifty device called a "Punch Card"
None of this is NEW. It is all just a rehash of ways to make money when the only money to be made is with old ideas. Lets start charging people for time again...that will boost our revenue!
It sucked then...so it will suck now....
You keep going until you die..."Me".
> They can simply pay for CPU cycles just as they pay for electricity
Oh, so they don't need CPU cycles to connect to those rented cycles? They submit data and receive results by US Postal Service? Given that the equipment on customer premises is usually powerful enough for most typical processing required by your generic ACME anyway, what extra cycles exactly do they need? Today's company needs lots of STORAGE and NETWORK BANDWIDTH to store and transmit endless Outlook memos with large attachments, their processing demands are usually more than adequately met. And your typical ACME will be far too distrustful to let some faceless third party store those Outlook messages for them. Never mind that a big fat external WAN pipe to that third party is a lot more expensive than a big fat internal LAN pipe.
with a pre-determined rate, IBM would have to build a lot of capacity to satisfy the peak demand times, but then they would have all that power sitting around unused when demand dies down. people who choose to use cpu cycles during off hours will end up subsidizing the peak demand users.
if IBM charged based on current demand, they wouldn't have to build up as much capacity because people would smooth out their usage to take advantage of cheaper prices during odd hours. before you submit a job, you could check and see what the current rates are, and also look at historic rates to try and predict a cheaper time to run your job.
sounds a lot like the arguments for running electricity markets...
The revolution will be in revenue.
Currently IBM big customers buy a new machine every four years or so, they pay a yearly maintenance bill. IBM has trouble predicting it's revenue quarter to quarter, in a downturn everyone stops capital expenditure and IBM mainframe sales plummet.
Under this model everyone should pay less but they'll pay every month like clockwork.
Computer Associates has a similar scheme for software. You rent your software on a monthly basis.
On a technical level I'm all for it. I have a suite in my current site that is run yearly and takes for ever. Currently IBM has a big box sitting here and we just sip from it, until year end when we max it out for like two weeks. Let me rent time on a huge box and I'll be happy. Gurantee my data and response time and I'll be ecstatic.
It could work if they had the right implementation. I run a lot of MATLAB simulations that require a day or so of compute time and lots and lots of disk space. Would I pay for just the compute time? Disk space too? What about installing MATLAB? Do I get my own Virtual Machine?
They can either do a kick ass job, or they can screw it up and it'll go down the tubes. In the end all that matters is everyone gets their jobs done with less money had they not done the processing locally.
Bad news: the rates charged per byte/cycle/whatever ought to drop by 50% every generation (12-18 months these days).
More bad news: typical supercomputer code is usually bummed (at least a little) for the particular hardware it runs on, to get the last factor of two or so for performance. If you rent crunchons, can you afford to rent generic crunchons and give up that last bit of optimization?
Good news: if you can get around the bad news above, this could turn supercomputing into a lease-vs-buy situation, and when the computer you buy essentially depreciates 50% every generation, leasing might be a win.
To a Lisp hacker, XML is S-expressions in drag.
From what I recall, Enron sold electricity and got in trouble for cooking the books. That said, the electricity itself remains a *physical* commodity, and it is measured in real world physical units. A kilowatt of electricity is just as real as a gallon of gasoline.
Computing power can also be measured and sold in a similar way. That said, I'm not sure if I would like to have IBM be my CPU-cycle utility. I prefer to generate my own CPU-cycles using good old fashioned electricity (formerly sold by Enron) and silicon from several sources (Intel, nVidia, et al)
Well, sure, the Frinkiac-7 looks impressive [to student] Don't touch it! [back to class] But I predict that within 100 years computers will be twice as powerful, 10,000 times larger, and so expensive that only the five richest kings in Europe will own them.
--Prof. Frink
-- Is "Sig" copyrighted by www.sig.com?
Your message implies that this will be IBM's one and only method for providing service. It won't. You'll still be able to have a dedicated farm running your stuff if you want.
And how exactly can you compare this distributed approach to distributed.net? Makes no sense. Are you trolling?
Blar.
Mr. Uradu,
Allow me to say that you (and the other person that replied) seem to be a bit confused about two things:
1. There is no "faceless" technician at the other end. What IBM is trying to do is exactly the opposite of that. They want to put tons of resources behind their Grid Computing initative so that companies have good reasons to move to that environment (i.e.: less downtime, more secure data, better performance, etc). It is to IBM's best interest to provide the highest quality service possible, or else nobody will join in or everyone will drop out after their first year (and remember that this is a per-ussage service, so the happier the customers are, the better for IBM).
2. What consumes most of the bandwidth in an internal company network is actually "raw" data. This meaning, database calls, method calls, etc. However once you outsource your IT department, you'll simply use your applications (preferably) using a web browser. This means that the only bandwidth being used by your company will be to display web pages. All the heavy work will be done at the datacenter backend. As for memos and attachments, those are things that will depend on the situation of each company. Some might simply opt to deal with those directly in their intranets, others will outsource it.
Likewise you can get a machine with an big ol batch of CPUs, most of them disabled. Over, say, the Christmas rush you call your salesperson and have the other CPUs turned on for a month. Again: Strange but the corporate customers seem to like it.
CPUs cost real money up front to make, ship, install, etc. If they send me 20, I only use 5, how do the other 15 get paid for?
The only way I can think of is grossly overcharging me for the 5 I do use, or REALLY overcharging me for on-demand capacity.
(*nix bigots and such note: Yes, I know, your defined user space on foobox is restricted unless you've chmod'ed your ~ to 777 (which is of course bombastically stupid), but do keep in mind that a typical home luser is running Windows, and accordingly sees their computer as their ersatz "user space".)
This sig no verb.
test, ignore, new Slashdot server setup giving trouble
Scientific and engineering computing really needs this. Maintaining large server farms must surely be cheaper, for the same computing power, than keeping isolated supercomputers close to where they are needed. And as the amount of computing power on the grid increases, the time you have to wait for a computation to finish will decrease.
The problem, of course, is security. Companies would be very nervous about putting valuable data on a shared server. For example, the data required for a computing task might include the geometry of a machine part or the chemical structure of a drug. The algorithms used by the software being run might also be proprietary. It's even a problem in academic computing because researchers still patent their discoveries or compete to be the first to publish a particular result.
Just how much can you trust IBM? And if the grid becomes open, with multiple companies offering their computing resources, the problem is even worse.
I know people who have generators or windmills and are connected to the electrical grid. When power demands are high, the power company actually pays THEM for their surplus power.
If I have a nice Linux cluster that meets the "standards" for the grid (whatever they are), can I sell cycles back to the provider? Or is it just one way, in which case I'm trapped into doing whatever the grid wants me to do.
Our article today sounds like batch:
"computing power of a supercomputer for a short period" although they do go on to say "Other services could be delivered in much the same way".
This would (did) work in the early days of computing, when it was virtually unheard of for anyone except for large fortune 500 companies and the US gov't to have access to computing power.
Why would anyone be tempted to return back to this model? How many sub $500 or even sub $200 dollar computers, will it take for IBM to realize computing power isn't rare or expensive?
And if a company or organizaion needs incredibly massive computing power is needed then can turn to companies like this to provide the solution, again using cheap generic pcs.
To some it all up this is stupid, and now Palmisano looks like another idiotic buzzword chanting CEO. This will be yet another blow to IBM, and it will soon (IMHO) join the growing stable of companies (Compaq, HP and the "new" Cisco) that have been screwed by a clueless greedy CEOs. Somebody needs to cancel his subscription to Business 2.0
... A cluster of these super-computer centers?
During the cold war, the CIA uses IBM to provide cover for operatives. IN exchange, IBM gets access to intelligence relating to competition.
Fast forward to today. Dozens of high quality encryption schemes foil the CIA's spying. What to do? Their friends at IBM can help again: create a new paradigm that leaves IBM in charge of all corporate data security.
</rant>
- None can love freedom heartily, but good men; the rest love not freedom, but license. -- John Milton
You mean they actually build in the processors and make them just sit there until you pay more? So you already paid for all their manufacturing costs and everything, just not for permission to use them.
Well if there was ever any doubt in my mind, this clinches it. Business is absolutely fucking evil.
In some organizations and especially in Canadian Universities where the funding simply isn't there to establish a supercomputing infrastructure (an article last week referred to a University in the Southern USA that had greater supercomputing capacity than all of Canada) the ability to purchase computing power on demand will probably be of great interest to Academics and medium to small sized businesses.
But the money for IBM wouldn't just be in the sharing of the computing power.
Say that Company A lacks the programming staff to develop a program for assessing the success of their Oil drilling or other geological analysis they need to do. What if IBM developed the custom program on behalf of the company and as well sell them the CPU time to run it. Would be a great way to increase their Consulting business and this one simultaneously.
I think it will be a very successful business model.
I'm also curious if it might on the whole reduce the power consumption of Server-Farms as fewer would be needed if IBM had a number of them prebuilt and ready to serve up power. I would imagine that a number of Hollywood studios might prefer this model to constantly upgrading their own server farms.
As for implementation of the communication end of things something like CANET:
CANET3 would probably fit the bill.
> There is no "faceless" technician at the other end
There is no fire-able individual that gets a performance review from the company. If they're unhappy with the outsourced datacenter performance, they have only two recourses: cancel the contract or sue, and I assume that contract agreements would most likely preclude the latter. The human element is completely being overlooked in these equations. Managers like pulling their staff together into a conference room and whipping their butts in times of crisis, making them feel in control of the company. Outsourcing precludes that. Sure, it will be (and has been) tried anyway, and will (and pretty much has in the case of ASPs) fail. But be my guest.
> What consumes most of the bandwidth in an internal company network is actually "raw" data.
[...]
> This means that the only bandwidth being used by your company will be to display web pages.
Hmm? Database queries are actually quite network efficient and in many respects very similar to HTTP. You send a query and get back a recordset. If you used a thin client instead, most of the information inside the recordset would likely travel across the network anyway, only in the form of more bloated ASCII-inside-HTML (to be displayed say inside an HTML table). And if the web server and database server don't reside on the same machine, you'd actually DOUBLE the network traffic.
Many cases can be made for browser-based thin client computing, but reduced network traffic definitely isn't one of them. There's nothing network efficient about stateless gobs of ASCII and graphics.
Another thing is that, as you mentioned, the ASP model is mainly suitable for web applications. Unfortunatly, that is still not the majority of applications in many corporations. There are still no satisfactory web versions of office applications, and there probably never will, because they're intrinsically client-side; if you insist on serving them via a browser, they will still end up mostly executing code (ActiveX, Java, JavaScript etc.) on the client side, but inside a sandbox, adding much headache and little benefit (think saving and printing).
As long as they keep it within reasonabl bounds.
People do not want "shared computing"; they do not want to put their data on "borrowed computers" nor do everything on "rented computing power" or "rented space". IBM should realize that most people will still want their applications and most of their processes and files on their own computers.
What IBM should be offering -- and what it seems like they're offering -- is loaning supercomputer time to people (for a price) for specific tasks which they can't accomplish in a reasonable amount of time on their own computers. This is a reasonable and useful idea; however, it is hardly new at all. At the University of Rochester, there are shared computers within biology labs, where people dump some heavy-duty computing operations and pick them up later. This went on during the 60's when computers were so expensive no-one could afford them. In short, this is hardly new nor revolutionary, though IBM may be putting a new twist on it by trying to use it as a business model.
It makes sense. After all, most people don't need supercomputing power for the majority of their tasks; why spend money on a supercomputer when it'll be unutilized 90% of the time? But what IBM can do is maximize supercomputer utilization by selling a percentage of its resources to various customers; these customers save money because they pay on a per-need basis.
For example, I often run Bayesian phylogenies. Recently, I ran a Bayesian phylogeny with about 50 taxa in it. This took 7 days on a dual G4 (2x 800MHz) Mac. That's with all of the computer's power focusing just on that. The time requires to complete the trees increases at a steep rate as one adds more taxa. If I was doing 200 taxa, it would have taken two or three months.
So this can offer a great service to many people.
social sciences can never use experience to verify their statemen
Just like wordprocessors never defeated the ballpoint pen. You certainly can't write FUCK THA POLEECE on the side of a bathroom stall with Word 2000.
Perhaps they're referring to a focus on marketing this idea to people that may not know that one can do this sort of thing? I don't know, it's really a fluff piece anyway.
What we call folk wisdom is often no more than a kind of expedient stupidity.-Edward Abbey
Not only that, some of the older boxen had a switch on them that you could set to put the box in a higher mode of operation. It cost more but you got more speed. Of course you had to involve the nasty suits in on the decision because it cost money. Was it worth the Pepto? I say no.
:-P
If you really need super computing time, just create a BrandX@home screen saver...
--Peter
Disclaimer: I'm a member of the Blue Collective.
It's called Capacity Upgrade on Demand (CUoD). Check IBM's site for more info.
The machines generally have all their processor slots populated, and IBM can remotely toggle them on when you need more CPU power for a workload. This is a cost effective way of providing users with the ability to upgrade as their computing needs grow. The cost of the idle CPU's is marginal when the entire system cost is taken into account.
What's that? What prevents a user from toggling on an idle CPU themselves? Nothing...other than invalidating their multi-million dollar maintenance contract with IBM.
"I disapprove of what you say, but I will defend to the death your right to say it."
- Evelyn Beatrice Hall
Lots of companies out there do subcontract computing time. In a lot of cases, it would only make economic sense. Why pay for a supercomputer when you only need 1 month a year of time?
:P
Otoh, it might still be cheaper for some companies to buy super-comps or build beowulf clusters if they have so much computing that they'd be using the IBM service 24x7.
Finaly some companies might not want 3rd parties to see the data they're working on
autopr0n is like, down and stuff.
IBM launched this initiative with a full-page ad in the new york times for a "business time machine" to allow you to go back in time and fix business mistakes. Here's a story on the ad. It was supposedly produced by bagotronics; the website now goes to the page on ibm's site (the picture of the device is at the top right). Here's a reuters article on the ad.
Ok, it's not a dot com superbowl ad, but still clever. A subtle way to acknowledge that their new initiative is like the old mainframe days.
HIV Crosses Species Barrier... into Muppets
It seems to me that it's ALWAYS a bad idea to bet that top-of-the-line computing power by current standards is going to be a valuable commidity even next year.
How can IBM possibly expect to keep pace with distributed computing taking hold? Haven't they noticed that yesterday's supercomputers are getting clobbered by today's Beowulf clusters?
And further... what if the computing power increases in following Moore's law begins to exceed programmers' ability to waste CPU cycles? 5 years ago I knew my machine would be obsolete in 6 months... obsolete enough I'd have trouble running the latest stuff. Lately, I've been squeezing 2 years out a machine pretty easily and (XP aside) there's precious little software that I use other than games that truely requires top of the line software. Of course that doesn't mean that GM's crash modelling software doesn't need as much power as it can grab, but you don't have to look further than CGI movie rendering over the past 5 years to see how quickly power increases and how quickly cost decreases.
- StaticLimit
We're in the process of discussing a move to thin client. We have roughly 500 users. The maintenance costs of wandering around to 500 desks is tremndous. We're looking at going to a central cluster solution. Rather than each department buying a PC for their desks, we simply bill them for their CPU cycles. Those that use more cycles owe us more when it comes time to cluster in more machines. Cost effective in the long run, and a LOT less work for us admins.
I realize that's not what IBM is proposing here, but the idea of charging for CPU cycles is quite intriguing, especially when it comes time to upgrade your $40,000 box. If you can bill the users for their usage, your upgrade costs go down considerably.
There is no reasonable defense against an idiot with an agenda
:wq
The only time you'd have some real savings is if you have truly variable needs, which I guess would be some companies, but not that many. But they've tried this since the dawn of computers, and it's never really worked out.
;)
However, with todays networks I would think booting off a gigabit LAN is doable, so I can imagine companies changing to a server/dumb terminal sysmem (again), but I don't see that server as being run by IBM either.
Of course, to make any real headway you need to get users to accept that you don't need a PIV 3GHz+ to run Word (not yet, anyway)
Kjella
Live today, because you never know what tomorrow brings
I think most people are misunderstanding the kind of applications this is targetting. people who need a supercomputer for a definite computation can already buy CPU time.
What IBM is talking about here is the anti-slashdot effect. I no longer need to engineer my server to have it survive the peak demand of my customers. I put my application in a DB2 farm maintained by IBM, on routine days it shares a CPU with other customers, the one day where i am on slashdot and everybody wants to buy my product, IBM lets my application use more CPUs, I only get charged for a high volume server if and when i have high volume operations ( and hopefully high volume revenues).
Dev elpizw tipota, dev phoboumai tipota eimai lephteros http://euclidian.org
E.g. the example with the car-industry is wrong (I don't work in the car industry, but I can guess). By the time they are finished with their design, the group will move on to the next model, and they will still need their big iron. It's not like the researchers and engineers will first design the car, and then get transferred to the production halls where they will be producing it, untill management decides it's time to make the next model!
What's wrong with recursive loops ?
func int factorial( int x ) {
if x < 1 return 1;
return x * factorial ( x -1 );
}
Two wrongs may not make a right, but three
Yes this concept is Timesharing on Steroids, but check what this CEO guy has already done- sold the commodity hard-drive biz and gone for Linux in a big way. He is clearly not risk-averse and assuming we all agree Linux is A Good Thing (and certainly a way to beat on Sun and Microsoft) he is not stupid. So what is he doing here?
Posters who are focusing on the U-word (utility) need to see that IBM doesn't want Joe Citizen using this. The profit levels for dealing with the general public just aren't there for IBM- Big Blue is all about the corporate or government cash.
In a word, cost savings for premier customers, i.e. the kind of people who will run up huge MIPS but not on a constant daily basis. Scenarios that come to mind beyond the car engineering ones are banks/companies/bureaucracies who have monster End Of Month/End Of year processing but reduced needs otherwise, websites that have a lower average use threshold except when the Super Bowl commercial airs, and disaster recovery (keep your disks mirrored offsite, if a disaster occurs call IBM, get your virtual mainframe up and switch to the offsite array).
With IBM's sysplexing and workload algorithms in play, tying in 'outside' 'puters will waste few resources.
I suspect that IBM's ultimate goal is disk farms on user sites and CPUs at IBM's Grid Ranch. With the CPUs under IBM's care they can really drop the TCO for the machines themselves.
That reminds me, the real cost of operating mainframes nowadays beyond the staff is the third-party licenses for the support software- security, tape libraries, etc. That's because traditionally the software vendors license by MIPS on the machine, not MIPS actually used in your LPAR (logical partition, a carved out virtual machine on a mainframe). Whenever you increase the MIPS of your machine, the third-party vendors will bleed you dry (which ultimately loses IBM customers as they go to cheaper alternatives).
IBM is beating on these vendors by competing in their arena to drive TCO down, and is also trying to get them to meter their actual usage under z/OS. So this grid thing is just a logical extension of what they are trying to do to not get run over by Moore's Law and the cost of running The Big Box.
________________________________________ History Must Not Fall Into The Wrong Hands ___________________________________
This setup is similar to old-fashioned timesharing, but on a far more massive scale. This puts together CPU power, network bandwidth, and storage capacity on a huge, cross-machine basis. The cross-machine capbility is the real trick.
With the technology in the works for this, you could combine together a zillion powerful boxes for the CPU power. (This would be similar to current clustering, but far more dynamic.) You could then attach an half an exabyte of storage, all of which would be pooled together for on-the-fly allocation of any given size, without any regard to which box the storage was in or what OS it ran.
All these concepts have been done before for decades. (Mainframes have done dynamic workload allocation for about 20 years, and AS/400's have used consolidated storage pools for quite some time.) However, combining them to form one massive pile of IT power, with the reliability you expect from the telephone company is a completely new idea.
You need another 10 terabytes of storage for a database expansion? No problem! Punch a few buttons, put in your Visa number, and in a couple of seconds, your filesystem is now twice as big. Maybe 2 terabytes came from a data farm in Boulder, CO, 6 from Gaithersburg, MD, and the last 2 from RTP, NC. None of that is your concern.
This, my friends, is new.
Maybe it's just that I'm reading at +4, but I'm suprised that nobody's mentioned Multics yet. The first thing that popped into my mind when I read "computing at a public utility" was Multics. I mean, the whole point of the system was to make computing a metetered utility. Not that any significant conclusions can be drawn from this, since Multics' failure had nothing to do with the business model, but more to do with them having overly ambitious goals for the project.
my sig's at the bottom of the page.
A business will always choose a fixed cost over a variable cost. But there's many points of view.
From a system administrator's point of view:
I work in the data processing indstury, and we have a 12-way NUMA box as our mainframe. We moved from a 16-way SE-70 that we'd had for seven years earlier this year, and our software has already expanded to max out the capacity of the new NUMA unit - to the point where we've upgraded it several times.
We'd continue to expand if the perception was that we have unlimited resources.
From a business point of view:
Even if we could do our dp activities on someone else's mainframe, we would still have our system administration costs for systems that can't be moved out of the building, so our costs don't go down. We would also have to maintain in-house development machines, because we wouldn't want to pay someone else for the endless compiles that we would need while developing new software.
Additionally, we already have a huge, unamortized investment in fixed dp assets.
Currently, our systems process for 24 hours per day to meet our needs. If we were to do these same activities on a metered system, we would probably not have to process as long, but if costs are over $5,000 per month metered, it's not worth it, especially since there are no cost savings except for the cost of amoritization of our main hardware.
Corporations buy unmetered data lines because they don't want to have to deal with variable (and, in case of a slashdotting, extremely high and exteremely unstable) costs. Trying to sell a service that has a variable cost structure is good for a company, but buying a service that has a variable cost structure is bad for a company. The only time buying becomes good is when the company can't provide it for itself, as with electrical power and telecom. But it's easy to buy/build your own mainframe-class computer for less than $10,000.
--
Vote for your hopes, not for your fears - Vote Third Party
That's the model that commercial renderfarms have been using for years. Guess what - it only works up to a certain point. After that, it's more affordable to just own and operate the resource yourself.
The exact breakeven point depends on the rate charged, and utilitization. But it's definitely there. IBM's ignoring a lot of operational history and available market research, if they really think this is going to fly.
This is not just about paying the meter. It is about utilizing all the wasted CPU cycles.
And it is massively clustered parallel sysplex.
I came, I saw, I LPAR'd
All Hail z/OS !!
After reading the article from the link, I was miffed about what this "on-demand computing" was and if it was any different than what's available today.
;)
In the USA today (yep, consulting on the road) The explain it like so: (Using an ordering web site as the example)
1. Each division of the company has it's own computers
2. When the orders increase in unit X, the computers deflect some of the processing to the corporate data network.
3. The newtork finds computers in other parts of the company that have excess capacity at that moment and routes the processing to them.
Side notes:
A. If the unit needs more processing than it can find inside the company, the data can be sent to other computers anywhere in the world, be processed and sent back.
B. If the economy turns sour, the unit might cut staff and let and outsourcing company handle all the order processing. Because computers would run on common standards, the division could give its data to an outsourcing firm.
I'm still miffed myself. Will this save money for companies or provide value that is worth paying for? It comes down to this, HW/SW/People/Real-Estate/Power(Diesel Generators) costs money (that's why you don't build your own datacenter) and technolgy gets old fast. If you truley can harness the X% of "wasted" computer power, in a "utility" fashion, then a miracle has occured. The technology of all technologies XML must have been mentioned somewhere and I missed it.
Article also says "grid computing" is a pre-cursor to on-demand computing. Sounds like we go back to green screens and mainframes to me. Oh wait, isn't that what the web is, but with fancy windows?
I think that most businesses won't have much use for this or other supercomputing apps. Like it or not, we use servers mostly for things that would not work well in an arrangement like this.
That being said, I am sure there are lots of companies out there that have supercomputers which they aren't using all the time, and if this would be secure, and less expensive, they might be willing to go to this sort of system. It would provide more flexibility. And might revolutionize that market.
Of course, *that market* is a very small piece of the computing market...
LedgerSMB: Open source Accounting/ERP
I'm a canadian!
We have dollar coins here. For those of you who don't know, it's called a "Loonie" because the coin features a Loon on one side.
A few years ago we also introduced a two dollar coin, called the "Toonie", whose name is really more of a bad joke.
It is in my opinion that with all the money our government saves by replacing the paper currency with coins we should recieve a tax credit for the cost of having all our pockets reinforced to carry the extra weight.
Those coins are damn heavy when you've got 20 bucks in loonies, and now they're talking about making a 5 dollar coin.
ATTN: Paul Martin, Finance Minister.
NO MORE FREAKIN COINS!!!!
What? Me? Worry?
Okay, great. When do they start? I'll take 10^20 cycles (on a power machine) asap, and pay about $1000 for it, as it will save me months on my thesis, allowing me to use mathematica, instead of re-hacking everything in C (which brings it down to about 10^14 cycles, manageable on a home pc in a reasonable amount of time). IBM, you sell to students?
I've had this sig for three days.
The Washington Post article writes:
."
... then again, that's when I do most of my work ...). This would allow corporations to most fully utilize their computing capacity, minimizing idle time for computer processors. (Of course, for people who have been in the computer industry, this isn't a new idea ... we're just returning to our roots, in some sense).
m
"International Business Machines Corp. chief executive Samuel J. Palmisano said yesterday that his company is investing $10 billion in a business strategy aimed at getting corporate customers to pay for their computing power in much the way they now buy power from utilities: as they use it
Presumably, IBM plans to charge a higher rate during times when processing power is in high demand (regular business hours) and a lower rate during "off-peak" times (wee-hours of the day
This is not the way power is sold for most residential and commercial customers in the U.S. and abroad.
Electricity is currently sold at a fixed per-unit rate, regardless of when it is used. The cost of running that server farm during Noon on the hottest day of the year (when everyone is turning on their air conditioning) is the same as the cost of running it at 2:00 a.m. on a modest Spring evening.
Many experts have pointed out that it is this lack of "real-time" pricing of electricity that has been one of the major contributors to the recent electricity crisis in California: http://www.sen.ca.gov/sor/Energy/Realtimememo1.ht
Whether this strategy will succeed is questionable. People have forgotten that one of the failed bets that contributed to the Enron downfall was their investment in "bandwidth trading." In a market with ample capacity, there is very little incentive to trade. Unless there is a serious processor capacity constraint, I can't see how IBM will be able to make an effective market for processing time.
Not so, have you ever been to Montreal..
No matter how many articles I've read, it always amazes me how few Slashdotters read the article before they feel compelled to post their (usually misguided) opinion. I'm sure plenty do, but there sure are a lot who don't.
IBM is working on the commercialization of Distributed Computing (henceforth, DC). This effort has been around for a while (in a related area, called Grid Computing, which some people use interchangably with DC) in the form of the Globus project, amongst others.
The concept behind DC is essentially a next-gen timeshare-- a distributed timeshare with an abstration layer, if you will. Unlike traditional timeshare, you don't specify where your processing will occur. Unlike existing projects (like folding@home, dsitributed.net), DC doesn't require that you have a parallel, segmentable computing problem.
Let's say (in your best Police Squad voice) I'm a mechanical engineer who's designing a car engine with a few thousand parts. I want to run some simulations on my model to inspect heat flows, vibration, whatever. Car companies (or the little guy with a copy of Catilla and a great idea) don't necessarily have dedicated computing resources to run my simulation. So, until now, I had to band together with a bunch of other mechanical engineers with jobs similar to mine and try to justify a giant simulation node. Or, I might convince management to outsource the computation, requiring a bunch of red tape, NDAs, contracts, negotiation, etc.
Now consider IBM, one of the largest commercial web hosts. IBM maintains giant server farms to support these services. Consider the amount of excess processing capacity sitting in these server farms because (a) a lot of servers are spitting out static pages and (b) extra capacity necessary to cover peak loading for special events.
Expand this idea to include thousands of people who need computation power for discrete, isolated projects and thousands of companies with excess computational capacity. The consumers don't care precisely where or when their computations get completed, they only care that they get done in a "reasonable" amount of time. An intermediary, which it looks like IBM wants to be, can accept jobs from them, break them into as many pieces as they can, farm them out to whichever of their suppliers has excess capacity at any particular moment, combine the results, and return them to the customer.
Even more, IBM can charge more if you want a high priority on your computation or if your job is not symmetric and must be run on fewer nodes.
Actually, if you think about it, IBM is hurting their server sales by advancing this project. Right now, they sell a lot of excess capacity to companies to cover their peak loading. If companies can dynamically purchase exactly the amount of processing they need, that's money IBM's leaving on the table. Now, companies with high-availabity requirements will still purchase their own systems with enough extra capacity to cover their own needs. But, when they're not using that capacity, they'll sell it.
I think IBM saw that the train was leaving the station. They know this technology is coming. And they see that the chance to be the intermediary in this market is worth more than the money they'll lose in hardware sales. And, they know if they don't, someone else will.
Timesharing is 95% obsoleted by the availability of cheap PC's and networks. It makes sense to do without the big iron and rent it when you need it only if you don't need it often. But for the occasional need, the overhead of making the connection, establishing both business and data interchange relationships with the vendor, getting the data to and from the remote computer, and getting everything working smoothly is very ugly. If they are selling me the same generic service that they are selling to my competitor, that's no competitive advantage to me, and I surely don't want to let them know about all the specialized custom computing that I do that gives my firm a competitive advantage.
The irresistable combination of commercial supercomputing, time sharing, and web services -- if you've got a trifecta ticket in your pocket with those names on it, you're on your way.
Was way ahead of it's time. The world just wasn't ready for boxes that came in any other color than beige
http://pcblues.com - Digits and Wood
perhaps they should be rather investing 10 billion minus half a billion in the ultimate tivo device running linux and able to connect to newsgroups for a certain fee. If this thing would be the ultimate (note: still no tivo in Europe!) then they could use the spare processor time of the device with a certain agreement: ' say you want to have a free use of our tv-guide, you have to donate an amount of free cpu cycles to our project. This way is best for both sides.
Investing large amounts of money into what at best can be a short term (5-8 year) money making opperation just seems unsound.
Computers are increasing in processing, storage, and display power at a geometric rate. Bandwidth is hitting a wall now in price and performance but an increased and more efficent infrastructure is expected to fullfill the needs. So called 'theoretical limits' are often debunked or simply worked around.
That's not to say this isn't a good idea to make money, it's simply something that could be done and put together for a signifigatly lower cost and ultimatly fullfills only a short term need.
The Internet is generally stupid
(The previous AC posting was by me, submitted from a logged-in state and not checking the "anonymous" box. Posting still isn't working right. I keep losing login info since the transition to "brak.slashdot.org". Maybe after more DNS propagation... Or do I need to clear some cookies, or what?)
Today, IBM was forced to take a ten billion dollar charge on their pie in the sky scheme to charge people for computer time after two thirteen year old hackers released OPEN-SETY, which lets large corporations use their PC idletime to handle large computational tasks in the background approaching the computational speeds of even the fastest super computers.
.netscare, but was able to design the 2005 model lineup, calculate the last digit of pi, determine the meaning of life (13), as well as design a nifty new toilet for the IT executive washrooms, using nothing more than the computational time of their secretary's idle computers.
It's estiamted that GM alone not only saved 15 billion dollars by using open-sety instead of IBM's
There is *no* sure revenue, There is no sure revenue.
When will these companies learn? We don't want to pay to compute, we want to purchase useful computing tools. That's where the innovation comes from.
The day this works is the day the computer industry stagnates....
Blogging because I can...
This is pretty much as expected from the company who, when the first computers were built, predicted that the world would need no more than a total of 5. IBM has pretty much demonstrated an amazing lack of insight into what the customer wants; this is certainly a good example.
I'm an American. I love this country and the freedoms that we used to have.
imagine...a cpu powerplant
a large GRID of Deep Blues. that is definatly worth selling.
...now imagine a beowulf cluster of those.
CAn'T CompreHend SARcaSm?
I'm convinced - there is no need for more than 5 computers, worldwide. (Thomas J. Watson, Chairman IBM, 1943)
I mean, I know IBM believes in low risk, but sticking to a 60 year old business plan is really pushing the issue. ;-)
If you were blocking sigs, you wouldn't have to read this.
After all, every time Slashdot faithfully reports another 5% increase in CPU speed, don't we ask "What does Suzy Homemaker or Karl Cube need all that processing power for? 500Mhz (750Mhz/1Ghz/whatever) is more than adequate for 90% of computer users. Only Rodney Research actually needs more speed." ?
Well, here's the answer. Suzy and Karl never need to upgrade again (after Palladium anyway, spit, spit). Rodney types up his research applications on his 1Ghz PC, then when he wants to fold proteins or discover a cure for belly lint, he rents two minutes on a Terahertz cluster. Heck, if it's made accessible and cheap enough, me and thee might even consider paying fifty cents to get a five second kernel compile.
The only losers are the chip manufacturers, because their business models are based on selling us more processor power than we need 90% of the time. That's OK though, because based on past experience, Gary Gamer will always pay premium dollar to go 5% faster than his friends.
If you were blocking sigs, you wouldn't have to read this.
This is what life used to be like in large company networks in the 70's and 80's. Mainframe accounts had timeslice limits. This seems like a very nice opportunity for IBM to haul out some of their dusty timesharing packages, dust them off, repackage them and make millions reselling them to customers.
John Manley is hte Finance Minister. Paul Martin resigned in May 2002.
Nine out of ten bathroom musings are written in pen (the other 1 is evenly split between Sharpie markers and car keys.)
What we call folk wisdom is often no more than a kind of expedient stupidity.-Edward Abbey