* Computing resources made available to wherever they are needed the most
Continuing our series of newsletters based on a Server Computing 2004 presentation I gave at CD Expo’s Enterprise IT Week in Las Vegas last month.
Utility or on-demand computing is the dynamic provisioning of a server so it can handle application loads as a service. This lets IT managers set metrics on server utilization, availability and performance for different applications. It also lets IT integrate business requirements with the IT resources required to support those needs and involves dynamically managing assignments of those resources based on business priority.
For instance, a retailer like L.L. Bean could use on-demand computing to handle the cyclical spikes in business activity from events such as holidays. During these periods, the retailer sees huge increases in demand on its critical business systems; if IT fails to keep up with demand, the business could potentially lose millions of dollars in revenue.
On-demand computing would let the retailer’s infrastructure dynamically shift IT resources from less-critical systems to support the most business-critical services, and return those resources to the “pool” when they are no longer needed.
The key components include IT technologies such as storage, provisioning, performance and availability, configuration management, automation, and security. It is the combination of these technologies with IT and business policy, in a fully automated environment that completes the on-demand computing vision.
The shift to utility computing, however, while utopian, will depend on market adoption a number of things, including:
* More complex server workload and provisioning software.
* Distributed server architectures such as blade servers and modular servers such as IBM’s new 4-16 CPU Itanium 2.
* Server partitioning technologies that allow the segmentation of multiprocessor servers.
With on-demand computing, IT can be promoted as a service so individual departments or divisions can be charged back for use. This aligns IT with business requirements so that applications can have the proper capacity and computing resources they need and simplifies management.
Just last year, IBM said it would spend $10 billion on on-demand computing. On-demand computing consists of three concepts, says Jamie Gruener, senior analyst with the Yankee Group. It consists of the ability to bill users for consumption, capacity or number of transactions. It consists of the IT services such as storage, computing and applications and access to those applications. Last, it consists of an architecture that includes technologies such as blade servers, networked storage, consolidation, virtualization and grid computing.
It is driven not only by vendors, but by changes in IT organizations to be more efficient, justify costs and shift resources to a pay-as-you-go model, which can dynamically shift as user-demand ebbs and flows.
