The adaptive enterprise drawn from multiple threads

When it comes to manufacturing and supply chain, Hewlett-Packard (HP) is focused on its vision of the "adaptive enterprise," which uses virtualized hardware, utility computing, service-oriented architecture, and other innovations to generate big cost reductions and performance gains. But typically, the first step on the infrastructure-optimization journey is much more prosaic—usually starting with consolidation.

Take the case of Los Angeles-based Belkin Corp., a fast-growing, privately held electronics products maker. The company needed to gain scalability and flexibility to respond to change and support 30-percent annual growth. To get there, Belkin replaced a sprawling, underperforming Sun-based server environment that ran its enterprise applications with a virtualized environment built on HP 900 Superdomes. The significant availability and performance improvements came at a lower total cost of ownership and required two-thirds less floor space.

In some ways, this is a typical virtualization success story focused on high-volume transaction processing. But HP is betting the next frontier will be in product design and other areas that can benefit from increased flexibility at the application level.

Accessing increased computing power when it's needed is the key. In the past, shipping, installing, and configuring new application servers might take months. But today, with clustering, virtual machines, blade servers and dynamic partitioning, new apps can be ready to go in a matter of days. Plus, multiple applications—not to mention multiple operating systems—can run on the same server.

HP also enhanced OpenView, its suite of network management applications. Manufacturers like Porsche have long used OpenView to keep tabs on diverse global IT environments and to manage network nodes. With new modules in place, they also can use OpenView for compliance management, identity and access management, and—thanks to the 2005 acquisition of Peregrine Systems—asset management.