June 17, 2005 4:00 AM PDT

Will computing flow like electricity?

As a provocateur, he's very effective. But as a prognosticator, people are less convinced.

Business writer Nicholas G. Carr raised many hackles in the information technology industry when he published a piece titled "IT Doesn't Matter" in 2003.

His latest piece with a similarly extreme headline, "The End of Corporate Computing," reopens the discussion of utility computing--the notion that corporations subscribe to computing services over the Internet much as they purchase electricity.

News.context

What's new:
Nicholas Carr, the author of the provocative article "IT Doesn't Matter," published another article, "The End of Corporate Computing," that predicts a large-scale shift to utility-like computing services.

Bottom line:
Industry executives queried by CNET News.com agreed that the computer industry will move to more hosted services over time. However, they see limitations to hosting and disagreed with the notion that the computing industry will evolve much as electricity did a century ago.

More stories on this topic

Yet Carr's latest article, published earlier this spring, failed to spark much industry soul-searching or a heated debate on the future of corporate computing.

IT executives queried by CNET News.com agreed that hosted services, or utility computing, will become more common and that corporations will take advantage of new technologies, such as Web services, grid computing and virtualization, to lower computing costs.

However, few executives envision a whole-scale transition to utility computing, even in the far-off future. None appeared to buy into Carr's assertion that the balance of power in the computing world could shift dramatically from technology infrastructure providers to Internet companies, such as Google or hosting companies.

For example, Charles Giancarlo, the chief technology officer of Cisco, downplayed the importance of utility computing scenarios. Like many others, Giancarlo said hosted services will become more important in certain situations but utility computing services will not be the norm in three to five years.

"We think (utility computing) makes sense for some small and medium-size businesses. But for large businesses, the decision to host applications outside or inside of the network depends on many different factors, including cost and network efficiency," said Giancarlo. "Some of the largest companies can run their own applications much cheaper and more efficiently than any utility computing provider."

"Overall, Carr has taken a very interesting analogy with some truth to it to an implausible extreme."
--Eric Newcomer, CTO, Iona Technologies

Other executives said that Carr's prediction that utility computing will become the industry norm is predicated on improper assumptions about the complexity of computing or blind spots in his knowledge.

In particular, Carr downplays the competitive advantage that custom-built software applications can bring, compared to hosted offerings, said Eric Newcomer, chief technology officer at software maker Iona Technologies.

"Computers do not work without software. And unlike electricity or other raw technology, software is designed for direct human interaction," Newcomer said. "Overall, Carr has taken a very interesting analogy with some truth to it to an implausible extreme."

Meanwhile, readers of CNET News.com, who responded to a news story on Carr's "End of Corporate Computing" piece, voiced a mix of opinions.

"The notion of the computer as computing device has been obsoleted by the Internet. All of the real action these days is in using the computer as a *communications* device," wrote one reader.

Others said that utility computing has yet to prove indispensable to corporate customers.

"I'd suspect that IBM's old mainframe philosophy is behind this drive to utility computing. It's very easy to bill by the month and provide premium services by the hour," wrote one reader. However, he questioned the need for these services: "I'm not aware of any pressing need that can only be met by Utility Computing."

Data centers obsolete?
To frame his discussion in the "End of Corporate Computing," Carr uses the analogy of the electricity industry and its own development over a century ago.

Carr argues that corporate computing data centers are analogous to private generators, which were used in the early days of electricity. These power sources burned fuel to generate electricity for a single site, such as a department store or a wealthy person's home. (Tycoon J.P. Morgan was the first residential customer in New York City in the late 1800s.)

But private and small-scale power generators, which used direct current, were eventually displaced entirely by alternating current technology, which allowed utilities

CONTINUED:
Page 1 | 2

32 comments

Join the conversation!
Add your comment
Utility computing
Utility computing will only be as reliable as the network upon which it is built. What will happen when the network goes down? Will there be a disaster recovery plan? What about security? It is for these reasons that an IT staff is still needed. Until someone creates a self-sustaining network, utility computing will be a pipe dream.
Posted by (10 comments )
Reply Link Flag
Good point...
The author also seems to have ignored things like security, redundancy, competitive advantage and half a dozen other things.

If your idea of computing is two clerk punching data into an Exel spreadsheet, then utility computing will probably work well for you.

I suspect that covers about 20% of the buisness world. The rest are using IT much more strategically. It not only doesn't make sense to move to online utility computing for the other 80%, they would be giving away market edge by doing so!
Posted by (52 comments )
Link Flag
Utility computing
Utility computing will only be as reliable as the network upon which it is built. What will happen when the network goes down? Will there be a disaster recovery plan? What about security? It is for these reasons that an IT staff is still needed. Until someone creates a self-sustaining network, utility computing will be a pipe dream.
Posted by (10 comments )
Reply Link Flag
Good point...
The author also seems to have ignored things like security, redundancy, competitive advantage and half a dozen other things.

If your idea of computing is two clerk punching data into an Exel spreadsheet, then utility computing will probably work well for you.

I suspect that covers about 20% of the buisness world. The rest are using IT much more strategically. It not only doesn't make sense to move to online utility computing for the other 80%, they would be giving away market edge by doing so!
Posted by (52 comments )
Link Flag
He's not even right about power
It's funny that centralized power would be the chosen analogy given the fact that many energy experts believe that future energy will be generated right where it's needed and centralized power will be obsolete.
Posted by skane2600 (31 comments )
Reply Link Flag
He's not even right about power
It's funny that centralized power would be the chosen analogy given the fact that many energy experts believe that future energy will be generated right where it's needed and centralized power will be obsolete.
Posted by skane2600 (31 comments )
Reply Link Flag
Analogy breakdown.
A few differences between electricity and computing that test the limits of this analogy.

You cannot customize electricity as much as you can computing.

Your particular usage patterns in the way you use electricity do not give away as much proprietary information as do your particular usage patterns of computing resources.

When you use electricity, you do not have to share highly sensitive data with an outside party.

Electricity does not change on a frequent basis in a way that changes the possibilities in which it may be used.
Posted by mepp (2 comments )
Reply Link Flag
Analogy breakdown.
A few differences between electricity and computing that test the limits of this analogy.

You cannot customize electricity as much as you can computing.

Your particular usage patterns in the way you use electricity do not give away as much proprietary information as do your particular usage patterns of computing resources.

When you use electricity, you do not have to share highly sensitive data with an outside party.

Electricity does not change on a frequent basis in a way that changes the possibilities in which it may be used.
Posted by mepp (2 comments )
Reply Link Flag
Nice thought but a long way to go...
There are several problems with the utility computing concept.

1. Everybody wants their own custom application. There are many reasons for this but it boils down to nobody likes to make do with a "general" application because the people in charge of making the decision and those in charge of actually using the applications don't understand them. I've worked in several major IT departments and in each case there was much duplication between systems due to lack of knowledge of the existing infrastructure, internal politics and general resistance to making changes to procedures to accomadate a general solution.

2. Inertia - Corporate data centers are large organizations within a larger organization. They will be resistant to change because it would mean their jobs. Plus, there is the cost of changeover to utility computing. That means data conversion, QA, end-user retraining and then there is the entire physical apparatus of the data center. What happens to that? Nobody is going to be quick to jump on this bandwagon.

3. Security - I think this has been covered in other comments quite well. Businesses do not want to risk their data and are willing to pay more to store it themselves.

4. Outsourcing - Another name for utility computing is outsourcing. Basically instead of maintaining your own data infrastructure you'd be outsourcing your IT department. Then instead of X number of companies with varying sized IT departments we'd have X Number of companies soliciting Y number of IT providers.

One comment stated something about fungibility and that hardware was pretty much hardware. Having worked in IT departments and for a software vendor I'd have to disagree. Sure, you can get hardware to generally be identical but hardware is a collection of paperweights without software, be it system or application software. The hardware/software combo is complex enough that each installation is pretty much a unique entity. Very much hard work goes on in IT departments to standardize hardware and software throughout an enterprise environment and there are always those machines that are an exception. Same hardware, same software, same settings but somehow two different machines will peform differently.
Posted by BrenBart (7 comments )
Reply Link Flag
Nice thought but a long way to go...
There are several problems with the utility computing concept.

1. Everybody wants their own custom application. There are many reasons for this but it boils down to nobody likes to make do with a "general" application because the people in charge of making the decision and those in charge of actually using the applications don't understand them. I've worked in several major IT departments and in each case there was much duplication between systems due to lack of knowledge of the existing infrastructure, internal politics and general resistance to making changes to procedures to accomadate a general solution.

2. Inertia - Corporate data centers are large organizations within a larger organization. They will be resistant to change because it would mean their jobs. Plus, there is the cost of changeover to utility computing. That means data conversion, QA, end-user retraining and then there is the entire physical apparatus of the data center. What happens to that? Nobody is going to be quick to jump on this bandwagon.

3. Security - I think this has been covered in other comments quite well. Businesses do not want to risk their data and are willing to pay more to store it themselves.

4. Outsourcing - Another name for utility computing is outsourcing. Basically instead of maintaining your own data infrastructure you'd be outsourcing your IT department. Then instead of X number of companies with varying sized IT departments we'd have X Number of companies soliciting Y number of IT providers.

One comment stated something about fungibility and that hardware was pretty much hardware. Having worked in IT departments and for a software vendor I'd have to disagree. Sure, you can get hardware to generally be identical but hardware is a collection of paperweights without software, be it system or application software. The hardware/software combo is complex enough that each installation is pretty much a unique entity. Very much hard work goes on in IT departments to standardize hardware and software throughout an enterprise environment and there are always those machines that are an exception. Same hardware, same software, same settings but somehow two different machines will peform differently.
Posted by BrenBart (7 comments )
Reply Link Flag
Long ago, in a galaxy not far away...
Sounds exactly like what was predicted back in the early 70s re TIMESHARING. Computing power would become a utility and purchased on demand like electricity.
Posted by codesmith (11 comments )
Reply Link Flag
Long ago, in a galaxy not far away...
Sounds exactly like what was predicted back in the early 70s re TIMESHARING. Computing power would become a utility and purchased on demand like electricity.
Posted by codesmith (11 comments )
Reply Link Flag
Security
I think one of the biggest hurdles to Utility computing is
Security. The need for security will manifest itself in two ways:
1. A delay in transitioning over to Utility computing: Until
customers are convinced that their information will not be
compromised they will be reluctant to transfer critical/high
volume processing. One aspect of this could be the need for
closed networks. Electricity companies typically own their
distribution networks.
2. A different breed of players: Security companies could get a
foot in the door ( a small foot though) by offering security
solutions for utility computing over the internet.
Posted by (2 comments )
Reply Link Flag
Security
I think one of the biggest hurdles to Utility computing is
Security. The need for security will manifest itself in two ways:
1. A delay in transitioning over to Utility computing: Until
customers are convinced that their information will not be
compromised they will be reluctant to transfer critical/high
volume processing. One aspect of this could be the need for
closed networks. Electricity companies typically own their
distribution networks.
2. A different breed of players: Security companies could get a
foot in the door ( a small foot though) by offering security
solutions for utility computing over the internet.
Posted by (2 comments )
Reply Link Flag
Electricity is not like Data
Electricity is a simple, physical entity.
Corporate data is complex, somewhat arbitrary (in that the corporation defines it) and needs to be consistent across the enterprise.
Utility computing can only be of use where the data managed by the utility does not need to be integrated into, and consistent with, Corporate Data. How often does this happen? Not often.
In most medium to large corporations, it is the data and business processes embodied in the applications that matter, not the technology. That's why Carr is correct when he says that "IT Doesn't Matter" but incorrect when he claims that utility computing is the answer.
It may sometimes be the answer but not enough to empty the corporate IT department.
Posted by (2 comments )
Reply Link Flag
Data is electricity
What is data when it is stored on a computer? Magnetic marks when in storage and electricity when being processed. When it is being processed, it doesn't matter what it is, text, numbers, graphics, AI, sound, whatever, those are abstractions. Minor point, but I am bored.

I don't think that utility computing is ready for prime time, and I doubt it ever will be. You are still using the same amount of computing resources, so what was saved? Nothing really, except higher costs from markup and slower access and processing time.
Posted by Bill Dautrive (1179 comments )
Link Flag
Electricity is not like Data
Electricity is a simple, physical entity.
Corporate data is complex, somewhat arbitrary (in that the corporation defines it) and needs to be consistent across the enterprise.
Utility computing can only be of use where the data managed by the utility does not need to be integrated into, and consistent with, Corporate Data. How often does this happen? Not often.
In most medium to large corporations, it is the data and business processes embodied in the applications that matter, not the technology. That's why Carr is correct when he says that "IT Doesn't Matter" but incorrect when he claims that utility computing is the answer.
It may sometimes be the answer but not enough to empty the corporate IT department.
Posted by (2 comments )
Reply Link Flag
Data is electricity
What is data when it is stored on a computer? Magnetic marks when in storage and electricity when being processed. When it is being processed, it doesn't matter what it is, text, numbers, graphics, AI, sound, whatever, those are abstractions. Minor point, but I am bored.

I don't think that utility computing is ready for prime time, and I doubt it ever will be. You are still using the same amount of computing resources, so what was saved? Nothing really, except higher costs from markup and slower access and processing time.
Posted by Bill Dautrive (1179 comments )
Link Flag
Sun Grid - the closest contender?
I've been following utility computing efforts for some time. It seems that Sun's Grid is closest, with simple CPU/hour and GB/month prices of $1 each. The challenge is whether my chosen software will run on the Sun Grid? Today the answer is no, but maybe tomorrow?
Posted by hutchike (157 comments )
Reply Link Flag
Sun Grid - the closest contender?
I've been following utility computing efforts for some time. It seems that Sun's Grid is closest, with simple CPU/hour and GB/month prices of $1 each. The challenge is whether my chosen software will run on the Sun Grid? Today the answer is no, but maybe tomorrow?
Posted by hutchike (157 comments )
Reply Link Flag
Flawed thought
Utilities deliver homogeneous products, electricity, or gas, that
is absolutely identical at each site with no customization at all.

Utility computing is just another name for webhosting, app
hosting and other centralized computing attempts. Just like
those, there are many promises but it always boils down to one
thing, the costs only are cheaper if all use the same app or
service with little to no customization. If the thought is "if it isn't
MS office you don't need it" then utility computing has a chance
to be big. Other than that it will become a respectible
component of the IT future. It will not be the only "IT".
Posted by kulro (5 comments )
Reply Link Flag
Flawed thought
Utilities deliver homogeneous products, electricity, or gas, that
is absolutely identical at each site with no customization at all.

Utility computing is just another name for webhosting, app
hosting and other centralized computing attempts. Just like
those, there are many promises but it always boils down to one
thing, the costs only are cheaper if all use the same app or
service with little to no customization. If the thought is "if it isn't
MS office you don't need it" then utility computing has a chance
to be big. Other than that it will become a respectible
component of the IT future. It will not be the only "IT".
Posted by kulro (5 comments )
Reply Link Flag
Clearly the way forward
Companies large and small are going to move across to this model so they can forget about the IT and can concentrate on the core business, I think in the future companies will run on thin clients connecting to trusted online desktop services like <a class="jive-link-external" href="http://www.cosmopod.com" target="_newWindow">http://www.cosmopod.com</a> for all their computing needs.
Posted by iqula (59 comments )
Reply Link Flag
Clearly the way forward
Companies large and small are going to move across to this model so they can forget about the IT and can concentrate on the core business, I think in the future companies will run on thin clients connecting to trusted online desktop services like <a class="jive-link-external" href="http://www.cosmopod.com" target="_newWindow">http://www.cosmopod.com</a> for all their computing needs.
Posted by iqula (59 comments )
Reply Link Flag
 

Join the conversation

Add your comment

The posting of advertisements, profanity, or personal attacks is prohibited. Click here to review our Terms of Use.

What's Hot

Discussions

Shared

RSS Feeds

Add headlines from CNET News to your homepage or feedreader.