The Need for Speed

By Jodi Mardesich

Successful organizations make the right data instantly accessible to management, employees, customers, and partners. It's a challenging task, as corporate networks grow in size and complexity, with more computers, applications, and data to manage. The management of data storage alone is daunting. The amount of storage capacity is increasing almost 80 percent annually, according to International Data Corp. At the same time, IT staff is increasing by only 5 percent per year. The responsibility falls to the CIO to keep the expanding network running efficiently, managing the storage of more information per IT staffer.

In order to keep the proliferation of data from clogging the network, slowing performance and impacting company-wide productivity, CIOs should consider more than just data management. Delivering a responsive, resilient corporate infrastructure involves planning in several areas: determining the organization's hardware needs; software needs; and the architecture, or blueprint for the organization's network of computer systems. All the while, CIOs must set and adhere to policies for managing the changing network infrastructure.

Making networks more efficient

To effectively plan ways to improve performance, it is necessary to take a high-level view and scan the network, creating a map of current resources, their capacity, how they're used -- noting where the bottlenecks are. Network management tools can discover all connected devices and draw a map. Monitoring tools can then be used to gauge the flow of traffic, and volume of traffic. With this information, the CIO can plan strategies for improving network performance. These improvements fall into three main categories: hardware, software, and network architecture.

Hardware improvements

Computer processor speed is growing exponentially. Moore's Law has held pace, with the number of transistors per integrated circuit doubling every couple of years. Intel expects that to continue through the end of the decade. The increased capacity of processors translates to computers, network adapters, routers, and switches that can process more instructions more quickly, and handle more traffic. But upgrading computer hardware, with faster processors, more memory, and hard disk space, is costly. With IT budgets tight, CIOs must make strategic decisions on where new hardware will be installed: new computers on the desktop, faster network adapters in desktops and servers, higher capacity routers and switches in the data center, or faster pipes. Installing new hardware doesn't necessarily translate into immediate performance increases. CIOs should also consider software improvements and network architecture, as well as provision for the management of new resources.

Software choices

Hardware lays the foundation for communication within the company and with outside partners, but operating systems and applications ultimately direct traffic and communications between the people using the network. It is critical for CIOs to choose an operating system platform that can scale to accommodate growing business needs. It should also be reliable, serve as a platform for the applications needed to run the business, and not lock customers in to vendor-specific solutions.

The battle between back-end server operating systems between Microsoft Windows, Unix, and Linux will continue. In its report, "Gartner Predicts 2005, Servers and Storage," the Gartner Group predicted that the Linux operating system will offer greater innovation and functionality to Windows by 2010. "Linux is maturing at a faster rate than other operating systems," the report stated. Gartner recommends that organizations evaluate Linux and train IT employees to use Linux.

Before purchasing hardware, IT executives should analyze network usage to ascertain if applications need more bandwidth. Often, servers are underutilized. According to the Gartner Group, utilization of just five to ten percent of total server potential is common. Recently, the practice of software partitioning has matured and become more widely deployed. Partitioning software allows multiple applications to run on a single server computer, saving on the cost of separate servers for each application. Conversely, some transaction-intensive applications are designed to run on multiple processors to distribute the load. Partitioning simplifies the network infrastructure by consolidating applications onto fewer servers, with less management overhead. The benefit is efficiency more than performance, but if speed is the goal, partitioning is a solution. Distributed computing such as this increases performance by spreading the load among multiple computers, getting tasks accomplished more quickly by operating different threads in parallel.

Network architecture renovations

Increasing network speed and performance isn't just about adding more servers and backbone capacity to increase power and bandwidth. It's about knowing how resources are being used. Where does the traffic go? Where are the bottlenecks? Monitoring tools aid IT executives in making networks more efficient, thus increasing speed.

Savvy CIOs create a network blueprint that allows for growth, planning hardware additions, software additions, upgrades, patches, and fixes, so that the result is not a patchwork of mismatched elements but a cohesive, well-planned whole.

Thinking strategically

In all these considerations, CIOs should look beyond the technology at stake and take business needs into account. Today's CIOs are more than technologists. They're business strategists, and should focus on achieving business-oriented goals rather than technological ones. The evidence of their work should enable them to outperform the competition and improve performance, which should be reflected in corporate financials.

One method of improving network performance is putting into place a data archiving strategy. Information that is used less frequently can be stored offline, leading to more compact data warehouses. Information will be just as accessible, through logical links that point to the data. Data archiving is a critical component of the reporting necessary to be compliant with legislation such as Sarbanes-Oxley. Besides, it helps lower operating costs, since smaller data warehouses require less storage, less server capacity, and fewer staff to manage the archiving process.

For the next ten years, managing static data will be the greatest enterprise storage challenge, Gartner predicts. As a result of ever faster pipelines, wireless networks, e-mail, and e-commerce, more raw data -- data that is infrequently accessed or changed -- will be generated. To be compliant with legislation such as Sarbanes-Oxley, all data could be considered a legal document. Keeping the network responsive and not allowing increasing amounts of data to clog networks and slow down performance will be one of the CIO's most pressing concerns.

Jodi Mardesich writes about business and is a former staff writer for Fortune.