Powering business evolution: Utility computing spurs IT-business alignment

by Ben Tamblyn

September 28, 2004 —

In this interview, Sharon Green, Director of Utility Computing at EDS and Bill Mooz, Senior Director of Utility Computing at Sun discuss what is driving the utility computing market, how it differs from the ASP/MSP model, and what is next on the utility computing horizon.

Ben Tamblyn - UtilityComputing: Do you see the development of computing in terms of technological evolution or business evolution?

Bill Mooz - Sun: We view utility computing from four aspects. There is a technology aspect, a financial model and business aspect and, very importantly, a change management aspect. The processes you need are different, the expectations and way of doing things within the business will need to change very dramatically. I think that the first wave of utility computing, has focused primarily on the financial and business model aspect. The initial utility computing solutions that we have seen operate with a variable pricing model on top of a traditional service model and a traditional technological model. What we're now seeing is a move into the next phase where providers and customers are beginning to address this change model and the technology model with the goal of achieving greater efficiency.

The existing financial model delivers two key benefits:

-- a greater ability to align your IT expenditure with IT activity, which should hopefully correspond with the level of business activity.

-- The second thing that is obtainable but not always delivered through utility solutions is a more granular understanding for the business, as to what their existing IT costs are and how they will be impacted by moving to different financial models.

This provides all organizations with the potential to move from traditional block-billing -- a one-size-fits-all to a situation where the enterprise can understand the cost of adding new applications, or making a cost comparison depending as requirements fluctuate. However, unless you address some of the technological and service delivery aspects (e.g., autonomic capabilities and virtualization), the granular level of business understanding I have mentioned often does not become visible to the business decision makers.

Sharon Green - EDS: I think to a large extent utility computing needs to be business led. If you don't deliver the business results, regardless of what the technology is doing, you will never achieve the desired results that directly support the creation of competitive advantages and the delivery of overall business objectives. We have started with a pure financial model and are constantly adding things to our utility computing offering, and now we're looking at how we can deliver genuine business billing. For example, making sure we can charge on an investment in utility computing that directly affects the business, emphasizing variables such as the business activity enterprise customers are involved in.

UC: What do you see as the key market drivers for utility computing?

Mooz: I think that the key market drivers for utility computing are creating economic efficiency by reducing ongoing operational costs and improving levels of scalability and performance. If you look at typical procurement cycles for enterprises today, to move from the point in time when a business decision maker says he is ready to invest and deliver a new service, even when it involves adding capacity to an existing service, the time taken between making the decision and service deployment is significantly distant. It starts with the customer's procurement department: the order is eventually placed by the customer, received by the supplier, shipped to the customer site, and finally deployed. The procurement lifecycle takes a long time to complete and it is arguably getting longer. If under a utility computing model this process can be compressed and, in effect, instantly available once the decision is made to proceed, procurement lifecycles cannot only be reduced but customers can also benefit from improved process management. However, cost reduction will continue to be a key driver.

Green: I think the tie between utility computing and business advantage will grow ever closer. I think when utility computing models began to emerge drivers were primarily financial and technological; I think over the coming years greater emphasis will be placed upon the idea of creating business agility.

UC: I listened this morning to a presentation from Robin Barrett at American Express. I got the impression that what they have implemented thus far is in fact a traditional outsourcing model with flexible pricing. How does utility computing differ from traditional outsourcing models and how does it differ from the ASP/MSP model?

Mooz: I think that the biggest difference between outsourcing and utility computing is the creation of a pool of resources that can be shared among multiple different users. Truthfully, many of the existing utility computing deals combine a new variable cost model with traditional service and technology models. One of the things we have seen is that customers are beginning to make the leap to bring in a number of enabling technologies to allow for the creation of this pool of shared resources and evaluating the collective service models that make this possible.

Green: Moving beyond the financial re-engineering that categorized some of the early utility computing deals, we are starting to see the emergence of technology that assists in creating the idea of a true utility model.

UC: So have we learned the lesson from four years ago or will history repeat itself?

Mooz: I think that an economic upturn will accelerate the development of the utility computing model. A prime candidate for a utility computing model is a start-up because utilization rates are relatively unknown. Start-ups generally do not know how quickly they are going to need to grow; they are unable to predict future economic climates and financial limitations on large capital expenditures. I think this will result in the re-emergence of ASPs/MSPs rather than people investing in a lot of their own infrastructure.

Green: I think if it were to happen again the industry would be better prepared. We have learned, grown, and matured as a result, so that if the industry were confronted with the same issues again we would be able to respond and move ahead much quicker.

UC: Why do you think the take up rate for utility computing has been slow so far? Are the business and technological benefits strongly aligned?

Green: I think there is a variety of factors that have affected the take-up rate. Firstly, the demise of the dotcoms has lead to customers spending more time assessing and evaluating their existing and future IT investment strategies. We are also seeing a period of what I would describe as technology catch-up. The technological advancement in the last 10 years has been so rapid that customers are trying to catch their breath, and are now looking at how this new model of IT investment, service delivery and consumption can deliver business objectives. Thirdly, an increase in customer maturity and understanding and an increasing vendor focus on delivering business benefits has provided the vehicle for a shift in customer expectations about how investments in technology are made. The simultaneous evolution of these factors will result in an increasing uptake over the next couple of years.

Mooz: There is also a fair amount of cultural change involved in a migration to utility computing in much the same way as there is in a movement towards outsourcing. That will take some time to work through. During the late 1990's when the industry was at its economic peak, the main focus was solving Y2K issues. Customers were more or less unaware of many of the underlying issues, but remained consistent in the spending patterns to ensure business continuity. When the down-turn came, a number of the inefficiencies of previous IT investment models were exposed, and what has followed has been a period of consolidation. What we are now seeing is the acceptance of new models. In the last two years there has been a noticeable shift from financial utility models to more comprehensive utility models that combine both technology, the business implications and of course the impact on change -- both structural and behavioral. What has emerged is a greater level of sophistication from all players within the industry -- customers, integrators and vendors -- to determine exactly what will impact operational costs, and what implications this may have for long-term strategic decision making.

UC: A number of industry analysts have suggested that vendors such as yourselves have relied on the technology driving your business, and suggest that on-demand will require you to restructure operations so that technologies are aligned more with service delivery capabilities. Do you agree?

Mooz: One of the deliverables we are seeing already is the ability to take advantage of scale, providing the ability to exploit economic efficiencies. I don't think at this stage everything is going to move towards an environment where computing is offered in a large, homogeneous public utility. This probably represents a vision of the future, and whilst some aspects of data center management can be well supported in this kind of environment others will be supported within the enterprise client. The implications from a technological viewpoint are that the design of existing systems will need to operate as part of a pool of resources that run multiple services. Through the creation of utility optimized systems, Sun is able to scale at the server level both horizontally and vertically. Server operating systems will need to enable the creation of either server clusters or optimally sized virtual systems that can be adjusted over time and in line with demand from the business. To achieve this you will need very high levels of security, because multiple clients will be running on the same environment. At the middleware level, applications will need to be integrated with the rest of the software stack to enable greater efficiencies in configuration management, emphasizing the management aspect of service delivery. Virtualization technologies will become increasingly important and I also expect to see increasing focus on open standards. Web services, for example, clearly supports this environment. On top of this technology model sits the business model. All components of the stack are going to need to be constructed to support both short-term and long-term business objectives.

Green: If you look at the shift we've seen in the last 5-6 years, technology providers have traditionally talked about speeds and feeds, now while this is still mentioned today, most providers lead with reducing or controlling IT budgets and the implications for productivity, and sales volume. I think this emphasizes the shift that is occurring. It is not all just about the technology, it's about quantifying the direct impact upon the business or customers won't invest.

UC: Some people view utility computing with scepticism because there is so much variance of opinion in terms of both the business and technological benefits. When you talk to customers, what are the typical objections they raise to utility computing?

Green: The client's biggest fear at the moment is around pay-per-use. The down-side from the customer point of view is a lack of knowledge about billing. It is important to remember that the vendors are in a similar position with regards to defining the billing process. Pay-per-use models are evolutionary in their composition and will change over time, as customers and vendors become more accustomed to these types of models.

UC: Is that made more difficult due to the fact that enterprise customers are generally unaware of potential total cost of ownership (TCO) savings unless they have a very clear idea of their IT costs?

Mooz: One of the key benefits that is emerging from some of the existing financial models is identifying and managing more efficiently some of the cost related data surrounding technology investment. This will be a necessary part of the utility computing growth cycle. Over time people will need to develop a greater understanding at a sufficiently granular level to understand exactly what their costs really are.

UC: Many industry analysts have suggested that enterprises may not yet be ready for a completely variable cost/investment model. Do you agree? What role are Sun/EDS playing in assisting customers to adopt this model?

Green: We're working very closely with all our customers to assist them in adopting this type of model. It is all about being able to come in and look at the customer environment, identify strengths and areas of improvement and what will then be required, to hold their hands through this transformation. It's a multi-year journey for most customers. Unfortunately utility computing is not something that can be just turned on tomorrow. One of the true advantages we can bring to the table is experience. We are able to give customers an understanding of where they are, a blueprint for conversion, and also provide an understanding of the expected cost savings inherent by investing in utility computing. Most customers are looking for the vendors to provide cost savings of x%. However, this is really determined by the starting point. Customers who have made significant attempts to create a standardized operating environment will reap the savings much quicker than an enterprise that is still operating in a chaotic mode.

Mooz: Two weeks ago we were asked by EDS to put together a cost comparison for a customer that looked at technology investment for a specific project under a traditional model and also under the utility computing model. We were also asked to identify how each differed in terms of delivery, management, and deployment. What we're trying to do is present these solutions to customers to ensure they have the data at both the financial and business level to make an informed decision. The other key point, as mentioned by Sharon, is that cost savings are completely dependent upon past behaviors. If we were to compare some of the recent deals we have done, the most successful customers have been the ones that have considered the financial model and its implications last. The starting point is determined by the requirements of getting existing architectures to a level that will support the eventual transformation to utility computing. This should essentially help to drive the core IT costs down. The key question then becomes identifying the model that technically and financially enables customers to align the transformation with the creation of business advantage and allows for growth and also the peaks and valleys inherent in any business cycle.

Green: The point of consolidation is extremely important. We have spoken to many customers who are in the middle of consolidating their environment, and whilst they are realizing some TCO benefits, a lack of controls and processes places a glass ceiling on the true value of consolidation. As a result some customers are finding themselves in a constant cycle of environmental consolidation. One of the key things we bring to the table is that process discipline, the controls that will ensure that customers don't find themselves in the same position as they were two years earlier.

Mooz: The rapid change in technological advancement in terms of virtualization, autonomic capabilities, etc. means that without this process discipline many customers may even find themselves further behind. In the coming years we will see all enterprise customers looking at consolidating their existing data centers as these new capabilities become available, but if you don't have the processes in place, it will be impossible to keep up with the innovation rate. If competitors can, you will always be behind.

UC: Change management has become a new buzzword and seems closely linked to utility computing. What role are Sun/EDS playing in delivering change for your customers? What will the impact be on the IT department (i.e., new roles, responsibilities). How has this assisted existing utility computing customers in creating greater strategic agility?

Green: The role of systems integrators and partners is critical here. The implementation of utility computing undoubtedly requires significant management of change, however, even more important, is the governance that is required to make it happen. Different skill sets are required and therefore significant reskilling and training will be required to transition many of the traditional roles of application developers or network administrators to become business analysts and relationship managers. Outsourcing in the last couple of years has provided the vehicle to begin this process or competency restructure. A couple of years ago we developed a program to assist the transition of people from traditional IT roles into business oriented information and process managers that can assist more than ever in delivering business objectives and creating value for shareholders and stakeholders. Delivering this will assist in creating this idea of strategic agility.

UC: How does utility computing today differ from traditional utility services such as electricity, gas, telephony? Why is a utility based model applicable for the provisioning of IT services in the same way as other utility services?

Mooz: This idea of a traditional utility model provides for a commodity level service that is available at a fixed tariff, on demand, is possible with respect to a subset of services that will be easily commoditized. The big difference between electricity provisions, for example, you only really have 2 levels of service, you either have 110v and 220v. If your requirements exceed this it either needs to be custom produced or you are setting up your own generating facility. At some levels it will be possible to provide commodity level services. The ASP model, for example, is providing a commodity level service offering. Similar offering from payroll application providers are emerging because of the difficulty in justifying the purchase of an application that is run for maybe 1-2 days per month.

Green: The biggest issue here is agreeing on what constitutes a true utility model. Analogies have been made with electricity providers but no one is in agreement about where the provision of service begins and ends. Some people believe that the lamp is part of the socket and others believe that the responsibility of utility computing providers finishes with presentation of the service. Until, as an industry, we define where utility services begin and end we will never reach this idea of a true IT utility.

Mooz: I don't think anyone in the industry can really answer this question. If you consider the delivery of utility services at within the home environment where there are a number of services that are presented. At the most basic level we have the telephone; on top of that sits cable television. There is actually a big difference because by purchasing cable television you are actually purchasing the content whereas the telephone delivers the pipe. An ISP alternatively provides a combination of content, the pipe and a messaging service. I think similar models will be seen with utility computing. Some people will just want the raw computing utility, others will want a complete utility model, but both are still utility computing just in slightly different flavors.

UC: When do you think it will occur and how long do you realistically envisage it will take Sun/EDS to create economies of scale and provide utility computing profitably?

Mooz: The way I look at utility computing is that it is in fact just another form of network computing -- providing network access to servers, storage, applications. This is something we have been focused on for over 20 years. So in terms of our existing investment in utility computing, we invested in N1 a number of years ago, we are now investing in a few features in our new operating system that are useful for utility computing. Looking back over a 20 year history tends to lead to where we're going with utility computing, as opposed to being unique. If you consider it from a financial standpoint and consider what it takes for a utility computing engagement to be profitable for all parties engaged, I think the key point is that the ability to share economic rent between customers, integrators and vendors involves identifying a starting point where you can identify and fix a number of issues within the customer environment to generate some significant cost savings. In a utility environment, where you provide a shared data center for multiple customers to use securely, you have more scope for delivering these economic rents. However, utility computing also caries some risks, because under traditional models the customer has had to bear much of the risk, under a utility model vendors and systems integrators need to cover some of this risk. We are constantly identifying ways of reducing our own risks to ensure that we can maximize this idea of creating economic value that can be shared across the value chain.

Green: As utility computing develops, we are always in a position to integrate new components. The existing customers have made the decision to invest in this multi-year journey; everyone is receiving immediate benefits. As each key component is implemented they are able to see further cost savings and we are also able to identify immediate returns on the investment we have made, not only within individual customers but across the board.

UC: I get the impression that the existing partnership between Sun and EDS is dependent upon Sun providing for the technological framework, whilst it is the responsibility of EDS to deliver the implementation in line with business objectives. Why do you think that the partnership will provide for a sustainable model?

Mooz: Every provider has limited resources and R&D dollars, and it is a question of focusing on core competencies. To use an analogy with the airline industry, no one airline is responsible for operating the airline, doing the flight scheduling, maintenance, engine construction, the assembly, etc. This allows our customers to choose best-of-breed solutions and allow them to make business process decisions whilst providing them with the support and expertise that we have made a core competence.

Green: We believe that utilizing each other's strengths is a much stronger solution than trying to do it independently.

UtilityComputing.com