Enabling On-Demand Computing
By Amy D. Wohl*

IBM invited 46 analysts to its Boulder facility, where amid 505 acres of mile-high-plus Rocky Mountain splendor (and hundreds of prairie dogs), IBM has inaugurated its On-Demand Data Center service; it will be one of four in the U.S. (with others in Raleigh, St. Louis, and Southbury, CT) plus eight outside the U.S. So far Service Delivery Center West is supporting 98 contracts, totaling $22.8 Billion (but many of those are, of course, over multiple years). The statistics are impressive. The Data Center has 111,000 MIPS, 1,100 Terabytes of DASD, 19,000 distributed servers, and supports 450,000 seats. It is serviced by 300,000 square feet of raised floor.

A better question might be what does it do and for whom?

We spent a day and one-half talking about just that. IBM executives see their On-Demand offering as being a way to help customers meet the market challenges of continuous changes, rigorous competition, unrelenting financial pressure, and unpredictable threats (to security, market dominance, etc.). All of this happening in a market where customers need to become very responsive, able to focus on their business, avoid fixed costs where possible (to support that flexibility), and be resilient instead of vulnerable. All that while demanding that any new investment be essentially self-funded.

No wonder On-Demand computing which offers customers the ability to pay-as-you-go, with no need to make capital expenditures or to hire and manage additional skilled workers looks so inviting.

If this sounds too simple, it is. In fact, while some companies do prefer having IBM own and manage some or all of their computing assets, many customers prefer to continue to manage their own or even to ask IBM to install this kind of flexible technology internally, allowing a large customer to get better asset utilization, while still controlling his own Data Center. IBM estimates that On-Demand could offer an ROI advantage that moves utilization from the usual 10 to 20$ of processors and storage to about 50%. Moreover, aggressive consolidation plans can simplify the customer’s operating environment.

This is not about the customer throwing out his current non-IBM or heterogeneous system, but rather about selecting technology and standards that can be supported by an open standards configuration. IBM assumes that they will be supporting many vendors’ hardware and software for most customers.

IBM is hoping to go beyond this. IBM offers to all if its customers – not just to those who choose to buy the On-Demand service – their On Demand Operating Environment. This is a way of delivering a modern, highly integrated and automated computing environment. IBM says ODOE delivers a virtualized, open, integrated, autonomic environment, which supports business process transformation, on demand. It is intended to afford users automation, integration, and virtualization based on open standards.

This, of course, is the kind of new automated, managed system that each of the systems vendors – HP’s Adaptive Enterprise and Sun’s N-1, for example – is also talking about. HP also offers a utility-style service as well as a customer-controlled one. Sun prefers to offer hosted services only through their business partners.

IBM also spent some time trying to differentiate between the On-Demand hosted service offering and the Utility Data Center as well as the On Demand Operating Environment. Between these and e-Business On-Demand (which actually means using On-Demand tools and Business Process Transformation to get to a Real Time Business based on Real Time information), we clearly have managed to confuse nearly everyone. Too many demands on our poor brains.

Definitions

Some helpful hints for keeping things straight:

e-Business On Demand can be implemented using an On-Demand hosted environment or it can be implemented using traditional on-site data centers. It’s how you hook up business processes, information, and solutions that count, not how the computing system underlying it works.

You may have an On-Demand system hosted for you in an IBM Data Center or you may choose to have one on your own site, entirely under your own control – or you may choose various combinations thereof. It’s entirely up to what happens to suit your needs best.

An On Demand Operating Environment is not the same thing as the UMI environment you are likely to find in an IBM Hosted Data Center. That’s because some elements of ODOE are not yet ready for the Service Provider environment and IBM uses third party solutions instead. As the IBM ODOE pieces are available in Service Provider versions, this is likely to change.

Phew!

Finance Is Part Of The Secret

There are complex financial schemes underlying these offerings. IBM has its own Global Finances unit to offer a variety of standard and custom plans to its customers. So far, the Boulder Data Center offerings have definitely been custom one-off plans, negotiated individually. But they have some commonalities. All of them are based on the idea that it should cost of much to use the systems less than 100% of the time as it would to buy and own them. Typically, for reasonable lease commitments (which can be changed during their term), the customer agrees to pay a base payment for some base (minimum) usage and then a variable payment based on additional usage. The more predictable that usage can be, the lower the payment will be (that is, predictability carries a lower price). These schemes are available for both hosted and leased for on-site use systems.

You will be interested to learn that one reason IBM can offer such good prices is that they have another whole business – recycling systems that come back in from their leases, including PC’s. We’ll save that story for another day.

Grading IBM

I’d give the IBM folks clear marks for the state of the offering. Obviously, it’s there and it works. In fact, an internal IBM customer and external ones told us about just how they’re using IBM’s hosted services to provide virtual computing as an on-demand service in their business. IBM also offered a number of road maps of both “how we got here” and “where (and when) we’re going next.”

Those (and those of their competitors) will be part of the giant charts we’re getting ready to share with you. We’ve decided they are too complicated to put into the newsletter, so they’ll be up on our website, linked to appropriate narrative – or you can read all about them up on the web.

Soon.

*Reprinted with permission from Amy Wohl’s Opinions. Further copying, without permission of the copyright holder, is explicitly forbidden.

Go to top