Utility Computing: Where Is It At?

In Part 1 of our series focused on dispelling the myths surrounding data center virtualization, we addressed the basics of the technology and the important role virtualization will play in the evolution of the next-generation data center (see Keeping IT Simple).

Now, in Part 2, we will identify the current stage of development of the utility computing model namely, what level of technology is available and ready to be implemented today. There are three distinct stages that make up the present migration towards a utility computing model:

This article will address each of the three stages and clearly identify where we are today in terms of the technology and overall industry adoption.

First is the concept of virtualized computing. The first step in migrating towards utility computing is to begin deploying virtualized network devices to work in conjunction with your existing data center network infrastructure. As equipment needs to be replaced, virtualized network appliances allow IT managers to “turn up” new virtual services to handle the load, rather than deploying an entirely new device.

Virtualization is achieved by abstracting functionality and management away from physical implementation, thereby enabling companies to deploy one virtualized appliance to take the place of tens to hundreds of legacy devices, providing significant reductions in capital expenditure (capex).In the simplest terms, virtualization combines the best of both worlds: The efficiency of shared resources with the performance and security of purpose-built systems. As a result, virtualization optimizes data center application performance by pooling processing resources across all services and reducing the time it takes to launch new applications. This is possible because network administrators are able to configure, manage, and deploy multiple services from a single device.

The second step in the migration path toward utility computing is the concept of on-demand computing. This builds on the foundation of virtualized computing in which shared resources are distributed across a number of users for optimum performance and availability. The benefit of on-demand computing is operational expenditure (opex) reduction and increased flexibility through automation of the already virtualized infrastructure.

Automation creates a very flexible environment in which specific applications can be quickly delivered and provisioned on an as-needed basis. This means that services and applications can be provisioned remotely, which reduces the need to send network administrators to the data center to reconfigure appliances each time a user requests additional services. This translates to a significant reduction in opex.

In order to reach the third step in this migration path and achieve a true utility computing model, you must be able to make resources available to end users on an as-needed basis and, in an outsourced model, charge based on the usage of those resources.

Utility models can also be developed by corporations – often called a “private utility” model – and the charging can be done at a departmental level, rather than utilizing a fully outsourced model. To facilitate this level of fine-grained charging, a billing component is added to the already automated on-demand model, enabling outsourcing and a charge-back mechanism so that the economic benefits of utility computing can be realized.So, what stage of this migration are we currently in today? Most experts will agree that we are somewhere between the first and second stages in the migration path. Virtualization technologies are here today, and virtualized components can be purchased in all three areas of the data center, including computing, storage, and networking. On-demand computing is achievable today as well, though it is in its early stages. Utility computing, however, is still on the horizon.

In order to take full advantage of what is currently available, companies often choose to incrementally automate portions of their standing infrastructures, piece by piece, on the way to a fully automated, on-demand computing model. As each element of the data center is automated, the new challenge becomes management.

Until recently, each individual environment (computing, storage, and networking) required a separate management system. New management systems are being developed, however, that enable administrators to automate management across multiple types of virtualized components. These management systems are enabling true on-demand computing today and represent the next step in the migration path towards full utility computing.

— Dave Roberts, VP Strategy and Co-Founder, Inkra Networks