Although it's still very early on, outsourcing, software as a service, "pay as you go" and subscription licensing models point to a broad, long-term change in the technology market and the way IT services are purchased. The desire to better align business practices with IT deployment and to automate processes suggests that enterprises will change the way they buy IT services in the long term. Utility models imply several things: an ability to move from cost centers to computational utilities run on a commercial basis; links to external service providers for peak loads; the ability to move non-core assets off the balance sheet once IT resources are logically consolidated; and paying only for resources actually consumed.

Accessing compute resources and data without having to own the computers isn't a new idea. But accessing resources that are hosted or managed by third parties is only one part of the utility computing story. Many of the larger early enterprise adopters are developing in-house IT utilities -- with utility -style access and charge-back mechanisms. Some are additionally considering the use of external resources for specific projects, but in-house utilities are fundamentally where most of the action is today.

Nevertheless, a vision for the future of computing services based on the utility model is taking shape. On one side are ASPs, MSPs and outsourcing and hosting companies. On another are telcos and service providers, seeking roles as trusted providers of enterprise IT services. Meanwhile, the major IT vendors have redrawn their strategies to suggest that the ability to provide IT resources and services on an as-needed basis is the industry's future.

It's entirely reasonable -- given the loaded expectations being driven by IT- vendor marketing -- that enterprise early adopters should therefore want to procure and pay for their IT in different ways and indeed look forward to the day when the use of computers matches the ease of other everyday appliances and utilities.

From the enterprise customer's point of view, the business logic of metering usage is compelling: you only pay for what you use. This is brought sharply into view when considering that the use and procurement of computing services is increasingly driven by economies of scale and the effective utilization of resources. But moving from ownership to utility will still require some insight. Early adopters are going to want a way to determine the likely cost to them when the meter is turned on.

From a supplier perspective, development around utility concepts is being played out in two areas: public utilities (HP Flexible Computing Service, IBM On Demand Supercomputing Centers, Sun Grid Compute Utility) and technologies to support internally shared/datacenter utilities.

Role of Grids

Most IT vendors and enterprise early adopters that The 451 Group has spoken with see grids, and the attributes that give Grid computing its meaning -- virtualization, resource reallocation, automation and self-management -- as providing a technology underpinning for new kinds of IT procurement, delivery and usage models, the most evident of which are SOAs and utility models. As Grid computing becomes important, it may also become more transparent. By the time it is important, it will be called something else -- utility computing, SOA or datacenter automation, for example.

As far as utility models are concerned, there are a number of ways in which grids are being applied. Grids enable a new model of internal service provisioning, where the IT provider bills early adopters -- a company's own departments, and perhaps suppliers or customers, as well. Enterprise early adopters may be able to earn back some charges by making resources available for use on the grid. Because this activity is all happening over the Internet, these resources could just as well be supplied all or in part by an external hosting vendor or in some hybrid arrangement.

Key Drivers

For a number of enterprise early adopters that we have spoken with anecdotally, bare-metal resources are not what they are looking for. They are seeking providers that can supply capacity plus the application for a single price, and without the enterprise having to take care of the licensing. Fluent contracts for difference trading have been mentioned several times by early adopters. It is an approach already being offered by Savvis with Oracle instances, for example. Early indications show the following issues are key to users:

• Reducing staff costs and hassle -- For many users, reducing staff is the attraction of the outsourcing model. Deluxe Labs says a specialist can do things better and more cheaply, and its main cost savings have come from not having to hire database administrators.

• Flexibility -- The goal is the establishment of globally shared utility grids, based on a reference architecture, in which any application should be able to utilize spare capacity, regardless of geography.

• Scalability, particularly during peak periods -- For game developer Turbine, the utility model is currently attractive in two areas -- testing and first-week (peak) load. Across the gaming industry, the first week in the launch of a new game provides a surge in demand of typically 200 percent of the normal load.

• Doing new things -- Early adopters, often initially attracted by cost savings, quickly move to being driven by the ability to do new things. For global bank HSBC, it doesn't believe that using a Grid/utility model will reduce its TCO or enable it to save money -- but it will allow it to make more money. Utility capacity will enable companies like Offshore Hydrocarbon Mapping (OHM) -- an oil and gas services company -- to be more competitive by doing things that no one else is doing using resources it could not afford to have in-house -- and doing them more quickly.

Major Obstacles and Challenges

There is a wide variety of potential obstacles and challenges, the relative importance of which vary by enterprise organization. The following is a summary of the key ones:

• Security -- Across the board, security is the key emotional (cultural, organizational) and technology (compliance) barrier to using utility computing for banks.

• User resistance - -User resistance and reluctance comes in many forms, not least of which is the fear that utility computing will not provide as promised.

• Metering and billing -- The immaturity of management and billing technology is a key challenge for the buildout of utility models. There will need to be a framework for handling the contractual and reconciliation elements for the kind of utility services that can be easily created and dissolved before such services can be implemented in commercial situations.

• Software licensing -- The issue of software licensing has been well covered in previous 451 Grid Adoption Research Service (GARS) research. It is a key obstacle for some early adopters, varying by vertical market and ISV approach.

• Performance -- Turbine Inc. has evaluated offerings from specialist companies looking to provide utility-type services for gaming firms. It sees them as having three main weaknesses -- typically high prices, an inability to meet its specific needs and a limited set of software supported. In terms of the latter, Turbine believes that some vendors only offer support for one graphics package, which would mean that its special effects would be severely downgraded.

• Multiple vendors -- A key challenge for HSBC has been to streamline the number of suppliers it inherited from its acquired properties. It had a range of siloed grids and multiple vendors, so 18 months ago it put a stop to any further vendor proliferation and adopted a dual-vendor strategy, using Platform Computing and DataSynapse. It hopes that this stops further diversification of grid product use, reduces the need to redevelop products and allows it to focus on application development instead of infrastructure, as well as to standardize across lines of business, preserve development expertise and remove dependence on a single supplier.

• Skills shortage -- The key challenges for HSBC are: lack of developer and support expertise; testing and proving; the departure from conventional operational strategy; perceived loss of control by developers (each team currently runs its own grid -- grid hugging versus server hugging); application design and use of third-party packages; datacenter capacities and constraints; business reluctance for shared infrastructure; adoption of unfamiliar technologies; and compliance and regulatory issues.

Coming Up Next

In the next GARS Grid computing report (to be released in January), The 451 Group will look at the opportunity for Grid computing in the manufacturing and industrial sectors. Many Grid vendors reference the manufacturing industry as a key vertical market. Using case studies and vendor profiles, this report will provide an overview of Grid computing activities within this industry, examine the current use and future requirements of enterprise Grid customers in this space, and discuss the potential levels of adoption. Key early adopters range from the aerospace and automotive industries to design and testing firms.

For more information on this or any other reports in 451 GARS, please visit: www.the451group.com/intake/gridtoday-dec05.

About William Fellows

William Fellows is a principal analyst at New York-based The 451 Group -- an independent technology industry analyst company focused on the business of enterprise IT innovation.