The Promise of Utility Computing

By Tom Schmidt

Like previous computing models -- mainframe, client/server, the Internet -- utility computing has emerged as the next step in the evolution of IT. At its best, utility computing should provide IT services from a computing grid, similar in concept to an electric grid, from which companies can purchase just the amount of resources needed, and only for the time they need it.

The utility computing model is based on a shift away from technology for technology's sake, and a shift toward aligning technology with business requirements. This shift is already evident in many IT departments today, where there is a greater level of visibility into the costs of various IT operations. With accountability and charge-back capabilities, business practices and pricing models will evolve for the mutual benefit of both the IT department providing the service, and the business units using the service.

This article explores utility computing through software in three key areas: availability, performance, and automation.

Preventing data loss and downtime

Integral to any IT utility is the availability of data and applications; after all, an organization is dependent upon these elements in order to function. To ensure the highest levels of IT availability, an organization requires a range of software solutions that support a wide spectrum of operating systems, applications, databases, and hardware platforms and devices to fully protect a heterogeneous IT environments. High availability also means having an effective disaster recovery solution in place. These solutions can range from simple vaulting of backup tapes to full site failover using replication and clustering technology.

Gartner Inc. research shows that fewer than 30 percent of Fortune 2000 companies have invested in a full business continuance plan. The reason for this oversight may simply be that the technical challenges seem too daunting.

The first step in a disaster recovery plan is to build a data center that can continue operations in the face of hardware failure, power outages, or other common IT challenges. To accomplish this, hardware, data, and applications must be redundant, highly available, scalable, and secure. Clustering is one popular tactic for building resiliency into a data center.

The next step is to build another resilient data center at a separate location. True resiliency is achieved by ensuring that business-critical information resides in more than one place, and that the two locations are not susceptible to the same disaster.

Keeping performance levels high

Although availability is at the foundation of any utility computing strategy, it doesn't guarantee a usable system. The system can be available, but run very slowly. Application users will not tolerate painfully slow systems because it impacts their productivity. Fortunately, solutions exist today that allow IT departments to detect, diagnose, and correct performance problems before end users notice.

Today's performance solutions provide enterprises with:

• Optimized end-user response time
• Improved overall quality of service
• Automated discovery of performance degradation and rapid resolution

Because business applications are multi-tiered -- database servers, application servers, Web servers, networks, and clients -- application performance must take into account the complete environment. How well applications perform depends on how well those infrastructure components work together. Maintaining a consistently high performance requires an application performance management (APM) system that tracks the information flow across those tiers. Also, the APM system must accurately assess the quality of service received by each user, zero in on problem areas, and identify appropriate solutions.

Tracking key performance indicators from each tier means enterprises can take a snapshot of the user experience by measuring actual -- not simulated -- application response time. Effective transaction monitoring can mean the difference between keeping and losing millions of customers for companies that deploy Web-based applications. Because site users are likely to get frustrated and go elsewhere to buy if they don't get a quick response to online inquiries and purchase requests, it is essential that companies be able to find and fix performance bottlenecks before they affect critical business processes.

Maximizing your IT investment

Because more than half of a typical IT budget is spent on labor, it's smart business to make sure that the IT staff has the proper tools to be productive. Also, the more mundane tasks that can be automated, the more the IT staff can be refocused to more strategic work. In a utility computing approach, automation provides the means for efficiently provisioning and managing resources.

Today, tools are available that give enterprises server and storage provisioning, allocation, and tracking functionality. Such tools can identify and inventory all heterogeneous resources in the enterprise, then enable IT to automatically reallocate under-utilized servers to those areas in the enterprise where the need is greater. Similar functionality is available for storage resources.

Automating the process of server provisioning means discovering server hardware, installing operating systems, configuring applications, and modifying network settings on multiple servers simultaneously. Administrators can save images of server configurations to a central location and redeploy those images to available server resources as needed. In addition, they should be able to personalize those resources with the appropriate application and network settings -- eliminating configuration errors and shortening server configuration time.

Traditionally, deploying multiple servers is a time-intensive process. Not only must each server be manually configured with the correct operating system, applications, and network settings, but each of these functions is often performed by a different group. By automating the process, multiple servers can be configured automatically by one person without handoffs between groups.

Automation products also simplify the management of storage systems. By automating the routine, manual tasks typically associated with storage management, a single system administrator can manage more storage while providing higher levels of performance and availability.

In addition, storage automation products enable administrators to move data between different operating systems and storage arrays, balance I/O across multiple paths to improve performance, replicate data to remote sites for higher availability, and move out-of-date files to less expensive storage without changing the way users or applications access the files.

Conclusion

As a concept, utility computing holds great promise for IT administrators. To be accepted into the mainstream, however, utility computing must prove valuable to IT administrators at each step of its evolution. Only then will enterprises succeed in transforming their IT department from a cost center into a value center.

Tom Schmidt writes frequently about information security topics. He has more than 15 years' experience as a writer and editor in high-tech publishing.