Columns

THINKPIECE: Utility computing throttles app dev's value

• By Jeff Kaplan
• February 1, 2005

On-demand, utility computing has drawn considerable attention and spawned much debate. Although it’s hard to argue about its theoretical benefits, many application developers refuse to believe utility computing will become a reality, and thus downplay the implications it could have on development.

However, enterprise executives and end users are already embracing the concept and adopting early iterations of utility computing. For instance, they’re increasingly subscribing to software services such as Salesforce.com and managed security offerings.

Purists might suggest these subscription services are far from the elaborate, virtualized data center-oriented architectures that are necessary to support more powerful utility computing environments. For enterprise executives and end users, these on-demand subscription services represent practical, first-generation utility computing solutions that satisfy day-to-day business needs more effectively than traditional products.

Anyone who thinks Salesforce.com is a simple application that can’t scale, or a flash-in-the-pan solution for small and medium businesses, isn’t paying attention to the company’s growth. In the past year, Salesforce.com has experienced an 80 percent-plus increase in subscribers and even greater revenue growth. This growth isn’t coming at just the low end of the market, nor is it the result of giving away its service to gain market share. Salesforce.com is also quickly gaining acceptance in large-scale enterprises and generating healthy profits.

Salesforce.com’s success is also directly responsible for Siebel, PeopleSoft, Oracle and SAP adding on-demand, hosted services to their shrink-wrapped software portfolios. These major players are jumping on the subscription service bandwagon to resuscitate sales after experiencing a severe decline in demand for their traditional CRM, SFA and ERP products, which enterprises are happy to abandon after spending millions of dollars deploying them unsuccessfully.

The net result is that utility computing compounds the growing preference to buy, rather than build, important enterprise apps. Instead of continuing to invest in application development skills to develop proprietary programs, or customize already complex apps, executives and end users are willing to accept simpler software service solutions that satisfy basic business needs.

Growing interest in utility computing is the next step toward the commodization of IT and application development. As the offshoring of software jobs demonstrates, expert app dev skills are becoming a commodity. However, that doesn’t mean the development process has lost its value. Instead, app dev will shift its focus toward how IT can be designed and delivered as an on-demand service to meet business objectives.

Therefore, developers will need to adopt a greater service orientation to survive and succeed. This service orientation will combine the best-practice attributes promoted by the IT Infrastructure Library and the latest Web-based, app dev tools such as XML and Web services. The ITIL framework looks at the deployment of IT systems and software applications in terms of six service management stages: 1. Service creation 2. Provisioning 3. Monitoring 4. Problem resolution 5. Reporting 6. Billing

For application developers to remain relevant in the brave, new world of utility computing, they will have to follow the framework, which means using their project management and relationship management skills to better understand executive and end-user requirements and preferences so they can design applications that are easier to access and use.

To keep pace with these changing expectations, developers will also have to design applications that can be deployed more flexibly. IT will need to monitor and manage the applications more effectively to assure greater reliability, and meter and measure their use so enterprises can allocate costs based on how they use them.

The movement toward utility computing is real, and application developers will need to board the train before it leaves the station.