Students

The road to a mature network

Computer network problems cost American businesses $100 billion each year. While the underlying issues may be technologically complex, anyone who has ever fumed about a company-wide e-mail outage or an overloaded web site has an idea of the costs, says Kevin Dooley, a professor of supply chain management at the W. P. Carey School of Business. Dooley and his co-authors have just published "Network Maturity Model: An integrated process framework for the management, development and operation of high quality computer networks." The book outlines a descriptive framework that helps businesses assess the soundness of their networks and improve the veins that carry the lifeblood of companies in the information age.

2007-10-10

Computer network problems cost American businesses $100 billion each year. While the underlying issues may be technologically complex, the manifestations are often all too apparent and easy to understand. Anyone who has ever fumed about a company-wide e-mail outage or an overloaded web site has an idea of the costs, says Kevin Dooley, a professor of supply chain management at the W. P. Carey School of Business. Along with two colleagues in the NMM Partnership — William Baumann and James Fritsch — Dooley has just published "Network Maturity Model: An integrated process framework for the management, development and operation of high quality computer networks." The book outlines a descriptive framework that helps businesses assess the soundness of their networks and improve the veins that carry the lifeblood of companies in the information age. And, of course, these networks are only becoming more important and more complex. Dooley and his co-authors propose a new take on the work of Philip Crosby, reported in his seminal book, "Quality is Free" (first published in 1979). The Quality Management Maturity Grid that Crosby proposed in that book was adopted by the software industry in the 1980s and 1990s, just as the industry was growing in prominence and importance. There are other maturity models for the IT industry but they do not solely address network issues. "They tended to deal with the issues at a computer science level. In other words, they were looking at software issues and data issues and weren't really considering the network as a whole, which includes the physical infrastructure," says Dooley. Crosby asserted that a business process matures and as it does so, it goes through stages. Like Crosby's grid, the Network Maturity Model has five stages. Quantifying and standardizing these stages is overdue, according to Dooley. He hopes the book, geared towards IT professionals, will prove to be a roadmap for organizations that are reliant on their computer networks — that would encompass nearly every organization. He also hopes the best practices described will help organizations improve the quality of existing networks and design new high quality networks in the future. A staged approach Using the NMM's rubric, organizations can self-assess the maturity of their network management and place themselves along a continuum. In the first stage, the network and the people who run it have obvious problems. The network works — and often occasionally at that — because of the efforts of people that Dooley labels as "heroes." These people are not surprised when they have to spend late nights and weekends fixing broken bits and bytes. The network is put together ad hoc without much planning for the present and even less for the future, because thinking about tomorrow takes a back seat to putting out the fires that crop up daily. "Success is not part of the system," says Dooley about the NMM's first level. At the second level, standardization begins to creep into the management of the network. Basic users' protocols which are often quite inflexible have been implemented. These protocols may take the form of a policy (and the document that supports the policy) which delineates what can and cannot be included in a web site. A company, government agency or university may have scores of people authorized to upload web pages. A person uploading huge files or unapproved applications can create a bottleneck or security hole that later needs to be patched. The third level improves upon second-level standardization. Rather than clamp down the system, however, third-level standardization is more nuanced and allows for customization within the system. For example, organizations at the second level may outlaw all handheld devices to the network, but organizations at the third level have the capability and experience to make exceptions. Perhaps salespeople may be allowed to use Pocket PC devices to access their e-mail when traveling. "At level two, you might have key or major process areas that are structured and have some standardization. At level three, you're expecting that to be completely widespread," says Dooley. The Network Maturity Model's final two levels, four and five, address the incremental improvements that make a company's network function like a well oiled machine. Level four serves as a bridge to the ultimate goal of becoming a "level five" organization -- an achievement not necessarily mythical but certainly rare. Only 5 percent of organizations can claim the highest, self-actualized level, estimates Dooley. The biggest difference between level three and level four is goal setting. A level four company sets goals for its network -- perhaps striving for a miniscule "downtime" percentage. It collects data and manages the network using this feedback to reach its goals. Evidence of success at reaching goals may be the data generated by servers providing feedback on their own health, or it may be the number of network-related calls the help desk receives and resolves. At level five, the fine tuning facilitated by all of the data from level four has resulted in an organization characterized by deeply ingrained continuous improvement cycles. A level five organization may only have 5 percent waste in its system versus 40 percent in low maturity networks. The difference can reduce the cost of the system 20-25 percent. Dooley has consulted for a company that draws up a 5-10 page post-mortem report any time there is a significant network breakdown so that there is a clear record of what happened, why it happened and how it fix it. These reports become a knowledge base which further improves the network. A level five network is not just hard to achieve — it is also hard to maintain. Why? Surmounting the level five mountain takes concerted, capable employees — the kind that others want. "It's very hard for organizations to sustain level five performance, in part because the people who gain experience in such situations become highly attractive to be moved to other parts of the company or to be hired by competitors," says Dooley. A long but valuable road Moving from the helter-skelter environment of level one to the quiet hum of level five does not happen overnight. Dooley figures that moving up a single level can take 18 to 24 months. Within that time, an organization may have to undertake extensive planning, obtain buy-in from other departments and decision-makers, win a larger budget and then buy new hardware and software. Next it must train existing employees and hire new people and, not least of all, shift the organization's culture to be more appreciative of network quality. The journey from level one to level five can take five to eight years but it begins with a single step. At any stage, there is a danger that inertia will set in and organizations will lose their commitment to improve. The Network Maturity Model is part of the answer as it allows organizations to benchmark where they are along the maturity continuum, presenting goals (i.e. reaching the next level or maintaining the highest level) and the tools to get there. Yet, in the end, whether or not an organization takes the first step and continues down a path of gradual improvement is largely dependent on its people. If at the first level no one champions network health and finds a way to create some ease in the system — perhaps by hiring another manager to plan improvements while someone else keeps the duct-taped system running — the system will continue to limp along unimproved. "Most companies are only at a level two, and I would say that's because they're not purposefully seeking to improve their capabilities," says Dooley. It can be a hard sell. Network health is not the sexiest topic when an organization has to decide how to allocate its resources. The network, after all, does not produce revenue. But network missteps can have huge consequences that do affect the bottom line. As the book states: "Networks provide access to increasing amounts of data that satiates our ever increasing needs for information access However, this dependence on data networks creates risk; a typical network outage lasting an hour can mean the loss of hundreds of thousands or even millions of dollars in lost productivity and lost opportunities." Between 2 to 16 percent of lost annual revenues have been documented as the direct result of network problems. One little mention in the media can overwhelm a company's web servers, depriving it of hundreds, if not thousands, of orders. A study from Contingency Planning Research found that businesses such as brokerage houses would lose more than $5 million per hour if their networks went down. In a celebrated debacle last year, Airbus ran into significant delays in producing its much-hyped A380 after the company failed to standardize the version of its design software throughout its network — a lowly level two error within the NMM. Along with other delays, the error helped contribute to hundreds of millions of dollars for which the company is now liable because of late deliveries to customers. On the other hand, Dooley gives high marks to Google for its forward thinking in scouting out new locations to handle anticipated network growth — and even building near hydroelectric stations in the Pacific Northwest to be closer to the power needed for the hundreds of thousands of servers it will install. As with networks far smaller than Google's, the goal is the same: to anticipate and address future needs in order to help the network grow, and, in turn, help the organization blossom. After all, when it comes to networks, there's no room for immaturity. Bottom line

• While various maturity models have improved other facets of information technology, "Network Maturity Model" addresses how to assess and improve network quality.
• Networks may not be revenue producing but when they do not work, they can significantly affect bottom line earnings.
• The Network Maturity Model's five stages constitute a continuum from poor quality level one networks to level five organizations with very mature networks. Most companies do not get past level two and only 5 percent attain the highest level.
• Improving a computer network requires time and commitment on the part of an organization as it takes 18 to 24 months to move up one level and five to eight years to go from level one to level five.

KnowIT