03 February 2008

No 'tipping point' yet in PC alternatives, Intel says

Found an article on Server Based Comptuing that includes all the current buzzword technologies and the author has a well written opinion with numbers. What is almost unique is a posted comment about the article. The comment is as good as the article. Have posted both of them.

But the chip maker is planning for growth in demand for server-based systems and will have products for it

By Patrick Thibodeau and published by Computerworld

February 1, 2008 (Computerworld) Intel Corp. has released survey data it has gathered in an effort to find out what PC alternatives are winning the hearts and minds of its customers. The short answer from the survey is this: None of them.

Other than Terminal Services, which include Citrix Presentation Server and Microsoft Terminal Server, Intel said that users haven't anointed any other alternative -- including desktop virtualization, application and operating system streaming, or blade PCs -- as a clear favorite.

Finding out which technology will make the most inroads in the enterprise isn't an idle question for Intel, said Mike Ferron-Jones, Intel's manager of its emerging model program. "Knowing this is really important because it informs how we are going to build our products," he said. With desktop virtualization, for instance, the question is, "How do we optimize our server products for dishing out high volumes of client virtual machines to thin clients?"

Intel conducted online interviews with more than 700 IT managers at medium-to-large-size companies to gather its data. The respondents were qualified to ensure they had some decision-making roles, Ferron-Jones said.

The survey found, for instance, that 64% of the companies were using Terminal Services, a well-established technology, and were deploying them to about 26% of their clients. In two years, that will rise to about 34%.

But among the broader spectrum of alternatives, Intel said users have no clear favorite. For instance, 39% of the respondents said they had some current deployment of desktop virtualization, including installations such as testing and pilots. But the survey found that users were deploying desktop virtualization to only 8% of their clients.

Other models looked at include application streaming, with 30% of the users reporting some deployment; operating system streaming, 15%; and blade PCs, 26%.

But when asked what percentages of clients were using any of these models, as with desktop virtualization, the numbers were much smaller: applications streaming, 11%; operating system streaming, 3%; and blade PCs, 6%. But the survey points to increasing deployment in all of these technologies. As a percentage of client deployments, Intel's survey shows increases ranging from 50% to more than 100% over two years, although the overall percentage remained relatively low.

"What the data is showing is we haven't hit the tipping point on one particular model yet," Ferron-Jones said.

Bob O'Donnell, an analyst at IDC, said he wasn't surprised by Intel's findings. "I think they are correct that it's still too early to tell," he said.

"They are all valid technologies," O'Donnell said, and user choice may be based on which is best for the application load. As to why this is an important issue for Intel to explore, O'Donnell said, "I think they are a little nervous about it because it potentially changes their model" by moving to servers and potentially non-Intel-based thin clients.

Intel already has a good idea of what will be needed for these alternatives and has products in the pipeline, such as an upcoming microprocessor, Diamondville, which is intended for low-power devices, including thin clients, Ferron-Jones said. But the direction that users take matters, he said. For instance, if operating system streaming takes off, vendors will have to make sure they are building I/O subsystems "that can really pound out high volumes of OS images in short amounts of time," he said, such as when systems are booted up in the morning.

And here is the posted comment.

Have we gone full circle back to the 60's?
Submitted by Wisesooth on February 1, 2008 - 17:15.



When mainframe computers started interracting with dumb (and later, smart) terminals, the mainframe "timeshared" the terminal activity.

When the PC emerged as a viable alternative, the mainframes morphed from foreground timesharing and background batch-processing into file servers as people adopted the "distributed processing model." Thus began a tug of war between do-it-yourself end users and a centrally controlled IT department. At that time, they called themselves "Management Information Systems" or MIS. Their customers suggested that they add another "S" to their acronym because they missed the boat, but that is another story.

The "mainframe in a glass cage" morphed into a data center with racks of networked servers. The rise of the Internet, point-of-sale, collaboration, contact management, and other server-based applications expanded the role and complexity of the distributed model. Deployment, security, configuration control, and other issues made management of the infrastructure difficult and expensive.

As the relative speed and efficiency of hardware components increased, the processor was so fast, it needed work to do. Storage became so vast that it needed specialized gear to do that specific function. Servers are being consolidated using virtual management software because we now have processors and front-side bus architecture that is fast enough to make that happen.

The emerging "thin client" model gives control back to IT, simplifies deployment and configurtion control, and prevents rogue software from inhabiting client PCs. Of course, the end users have declared war on IT because they have lost their freedom to do their own thing.

Voice over IP and the thin client model has placed a heavy capacity burden on the network infrastructure. That area is in need of the most help and is the emerging "choke point."

Subscription licensing is going to increase the capacity burden on the Internet. This is another potential "choke point."

Sometimes, the IT architects adopt a new model without thinking the project through to its logical consequences. If they fail to do this, the results are predictable (all bad).

Agemain, the CEO of the Chase Manhattan Bank in the 1960s had these words of wisdom to share. He called these principles "ALOPTs" (Agemain's Laws of Prudent Thinking). Here are a selected few.
1. The surest way to make the new system cost more than the old is to pioneer the wrong thing.
2. The surest way to lose in poker or blow a computer budget is to throw good money after bad.
3. It is not against the law for a sales person to puff his/her product. Beware of sales people giving technical advice.
4. If a consultant knows only one vendor's product be assured that the consultant will recommend that vendor's product. If the only tool one knows how to use is a hammer, the solution for all problems looks like a nail.
5. Do not expect 10 technicians to fix a telephone in a telephone booth in one-tenth the time it takes one to do it.

Until the next post,

Steve