The company fit for the new millennium needs a strong frame to support its heart and muscleReader ROI -- In this examination of infrastructures built for corporate deftness, readers will learn - Technical considerations for tailoring infrastructure - How data is part of infrastructure - Options for organising personnel It is an enticing vision: the living company, one that creates and brings to market new products overnight, one that changes business processes or practices in a proverbial heartbeat to meet new market demands or ward off competitive challenges. Creating a corporate culture that can absorb and thrive in all that turbulence will be challenge enough. The sad truth, though, is that IT infrastructures -- networks, servers, applications -- are the real roadblocks keeping such daydreams from becoming reality at most companies. Computers are supposed to make business easier, but the way they are hardwired together often makes business flexibility much, much harder. The company that is truly "organic", adaptive or evolutionary -- heck, even just breathing -- in the next century will have built a corporate information backbone that helps foster agility rather than hinder it. From a technology standpoint, such infrastructures will not look the same for everyone. Some forward-thinking CIOs today envision fat network pipes and thin clients; others talk about message-oriented middleware and object request brokers; still others think of off-the-shelf components and Web-enabled-everything. Whatever the configuration, conforming the technology to the right business goals is paramount. "What's difficult is aligning infrastructure with business strategy," says Peter Weill, director of the Centre for Management of Information Technology at the University of Melbourne's business school in Australia. "The average life of a business strategy is less than 12 months, whereas an IT infrastructure typically has to last from 5 to 7 years. And some database infrastructures have to last 10 years or more." While business strategies also vary widely, popular themes emerge for competing in the coming years. Sugarplums commonly dancing in CIOs' heads include the ability to change business processes or at least add application functionality more or less instantly. And to give customers (whether external or internal) more or less instant access to information.
Technologies -- and the corporate organisational structures to support them -- are slowly evolving that will help today's companies take a few significant steps on the road to a living, adaptive infrastructure.
Get a Backbone
Some parts of the infrastructure equation are fairly straightforward. Two basics of IT are network bandwidth and processing power, and the way to an adaptive infrastructure is to get lots of each. In large companies, standardising disparate network technologies is a big step toward creating a more flexible, responsive infrastructure. Aerospace giant Lockheed Martin is three years into an enterprisewide standardisation initiative; acquisitions throughout the defence industry's consolidation period left Lockheed with around 75 "heritage" networks, a mishmash of topologies and technologies. Now in place is an enterprise asynchronous transfer mode (ATM) network that "gives four times the capacity at less cost", according to Paul Pelotte, vice president of distributed computing and telecommunications for Lockheed Martin Enterprise Information Systems (LMEIS) headquartered in Orlando, Florida. Fewer constraints on the network provide elbow room for adding users, traffic or new applications. In some instances sufficient bandwidth can replace processing power or at least relocate it. At Men's Wearhouse, a retailer of men's clothing headquartered in Houston, the vision of the future includes centralised data processing with thin-client machines in its 425-plus stores. Executive vice president Harry Levy is overseeing the rewrite of the company's core store system in Java; the existing crop of dumb terminal cash registers will be replaced with network computer-type devices, connected by dedicated network lines to the servers in Houston. The goal is to create an application that is centrally controlled and flexible enough to allow Men's Wearhouse to expand into other geographical markets or even into other lines of retail. Despite ever-falling PC prices, Levy opted against putting heavy horsepower on the countertops. "You don't want to put systems in your store that have expected moments of failure because the people in the store aren't equipped to deal with that. We don't have technical people, so we thought with PCs we were going to run into significant administrative and support problems," says Levy. For Men's Wearhouse, which has a traditionally lean IS operation, it makes sense to concentrate the processing in a place where it is relatively easy to keep an eye on the boxes and let bandwidth bring its power to the point of sale. On the other hand, analysts at GartnerGroup suggest that beefing up bandwidth alone won't be in a company's long-term interest. Even the widest of pipes will get clogged as usage continues to amplify. These analysts predict instead the use of "intelligent networks" based on directory service technology, which will be able to identify users wherever they are and prioritise bandwidth and power for the particular tasks they are executing. Choosing a standard network technology or processing platform is not necessarily simple, but the decision to do it is in many ways more important than the particular technology chosen. The bigger trick with infrastructure is how to pay for it. An "anticipatory" infrastructure by definition means building capability or capacity ahead of actual need.
Convincing business units to get out their wallets when there is no immediately definable payback is notoriously difficult. Tom Davenport, a professor of management information systems at Boston University School of Management and director of the Andersen Consulting Institute for Strategic Change, talks about "stealth infrastructure" as one solution IS organisations have adopted-building capacity over time by simply buying slightly more for each individual project than that particular project actually requires.
However, that practice may not be necessary for long, as Davenport says it is becoming more economical on the vendors' side to build the highest level of functionality into all their products and then enable or disable features via software at customer request. In that case, additional server or network capacity is essentially just a phone call away.
Application infrastructure can be more difficult than network infrastructure.
If a company wants to change its billing procedure, for example, how quickly can the financial software be changed to allow that? If a new customer service application is needed, will it integrate with the data formats and query facilities of the existing databases? Application infrastructure today is more often an inhibitor of change than an enabler. The common scenario: With the central IT budget under pressure, business units are more likely to take IT purchasing decisions under their own control, resulting in stovepiped applications and a melange of platforms and applications that have more communication problems than a bad marriage. The gridlock thus created makes it hard to deliver new applications, which in turn stifles the company's ability to implement new business processes. Currently there are two schools of thought about how to build an application infrastructure that can keep pace with business change. One is to buy everything, including all applications, off the shelf and let the software vendors worry about integrating. With enterprise resource planning (ERP) software, much of the application functionality is already built in, and changing processes is (at least in theory) a matter of flipping switches in the software set-up.
ERP software is the hottest thing going and "the very existence of ERP systems sort of implies that most people are not going to build their own applications anyway", says Davenport. ERP vendors, who already handle fundamental corporate functions like finance, manufacturing and human resources, are feverishly buying or building more and more functionality for their packages, including supply chain management, sales-force automation and a raft of other capabilities. The alternative school of thought is to build a completely independent layer of middleware to handle all communication among applications and data sources. Usually such a layer is based on object request broker (ORB) technology or message-oriented middleware (MOM), both of which are available in increasingly robust and successful commercial packages. For example, International Data Corporation (IDC) in the US (a sister company of IDG Communications, publisher of CIO), reports that the message-oriented middleware market grew by more than 60 per cent in 1997 and will hike up to nearly $US2.5 billion by 2002. The basic idea behind these technologies is that applications communicate with each other through a third party, a broker. Application A can work with Applications B, C and D -- all of which may reside on different platforms and communicate over different network protocols -- without needing a separate interface for each, providing each one has a single interface to the central broker.
As long as any new corporate application -- whether packaged or home-grown -- can communicate with the middleware, integration is relatively straightforward. Both approaches have their ardent supporters -- and detractors. The rap on the prefabricated ERP approach is that it puts all competitors on essentially the same ground. "I'm not a great fan of this lemming-like trend to ERP," says Weill. "SAP is mostly a defensive move -- it brings you up to industry standard, but it says you aren't going to use IT for competitive advantage," he says. In some industries (notably low-margin businesses), where IT is not a key differentiator among competitors, that makes perfect sense. In other arenas -- Weill cites the financial industry as an example -- it could be an enormous mistake. The middleware solutions, on the other hand, have been hyped for a number of years without breaking into the mainstream as quickly as promised; Davenport repeats a wry maxim, "If object-oriented were going to change our lives, it would have done so already." The truth, of course, is that each approach may be right for a different set of businesses.
The Heart of the Business
Of course all these applications need something to work with -- data. And while most people don't think of data as part of the infrastructure per se, that's a mistake, according to Laurence Bunin, CEO of the consultancy Handshake Dynamics LLC in New York City. "The key, fundamental thing companies are going to need is a centralised customer data set," says Bunin. If strategic relationships with customers will be the basis for competition in the coming years, each company needs to formalise the methods and tools with which that data is captured, centralised and analysed. One of Bunin's clients is a power company that wants to be able to centralise and analyse the power usage of major corporate customers, with an eye toward creating flexible and attractive pricing packages. Attaining that level of business flexibility, where prices and products can be individualised for each customer, requires walking before running. Bunin says his clients usually start treating data as infrastructure in a single departmental or business unit first. One group within the company builds an application that utilises customer data in a new way, and other groups, seeing the application's success, jump on board.
Data warehousing and data mining are current technologies for putting this data set in place, and the Web is gaining momentum as the method by which users access the data. Weill argues that the centralised customer data set is not for everyone: "If you take a company that has seven or eight business units that aren't particularly related, you wouldn't need it," he says. Interestingly, for those businesses that do need it, some pundits express concern that the rush to ERP may be a serious mistake, ERP being notoriously baulky in the area of making customer data accessible and digestible. While Davenport thinks this weakness will be resolved quickly, Bunin says, "Yes, but not in our lifetime." Human Powered Pelotte's ultimate goal is for technology to manage technology. In other words, the ideal infrastructure would essentially run itself with a minimum of human intervention. But while network and systems management tools grow ever more capable, it remains clear that the human element will dominate for a long time to come -- no network is adaptive and anticipatory when it is down. Thus the IS team itself is a crucial part of the adaptive infrastructure. No single organisational structure fits every IS department, but the theme among forward-looking IS groups is to find the right balance between decentralisation and centralisation. In any given year, many IS organisations seem to gravitate en masse toward one of these poles or the other. "I think there's as much fad-following going on in IT organisational structures as there is elsewhere in management," says Jerrold Grochow, vice president of the corporate technology group at American Management Systems in Fairfax, Virginia. United Press International (UPI) is a relatively small company that, owing to financial extremis, got the chance to remake its entire organisational structure from scratch over the past 18 months. Then-CIO Andrew Meldrum and then-CEO James Adams designed their ideal, ultimately flexible organisational structure with the goal of being able to create new information-based products -- news packages, photo clips and the like -- on the fly. IS employees at UPI are assigned to one of four "market teams", which are defined by the markets they serve and which comprise representatives from each of the traditional vertical functions (marketing, editorial, sales, finance, technology and others).
These teams have clearly mapped out their existing skill sets; when customers request a new product or service, the UPI teams can quickly determine whether they can meet the request. The IS group also meets separately from time to time in order to see what common requests for new skills are coming in. However, the group's first allegiance is to the market teams, not to the IS function. In larger corporations, centralisation seems better suited to manage resources and track projects and workflow. Lockheed's LMEIS is a separate company of more than 4000 employees serving Lockheed's numerous business units. Pelotte and company have created three Enterprise Service Centres, one at LMEIS headquarters, the others in Sunnyvale, California, and Denver, to "follow the sun", providing system support around the clock. In addition, most LMEIS workers are dispersed throughout Lockheed's business locations to handle problems onsite. The centralised service centres help Pelotte determine which services can be handled on an enterprisewide basis with concomitant economies of scale.
Despite all the organisational and technical progress, the visions of instant information and transformation are elusive sugarplums indeed -- the stuff of dreams for most companies today. As Davenport says, "I don't know anyone who has anything remotely resembling" those two capabilities right now. Still, some already achieve excellent payback for the time and money invested in building a better IT backbone. Weill's research finds that companies with excellent links between infrastructure capability and business goals deliver a premium of up to 40 per cent in business performance over the run-of-the-mill organisation.
About one company in five achieves that infrastructure-based performance boost, according to Weill. Making the right infrastructure decisions now, with the long term in mind, can help smart companies move into that elite group tomorrow.
RELATED BOOKS: Leveraging the New Infrastructure: How Market Leaders Capitalise on Information Technology Peter Weill and Marianne Broadbent, (Harvard Business School Press, 1998). The authors use case studies to support their argument that the best IT investments are managed with the precision of a financial portfolio. University of Melbourne (http://www.unimelb.edu.au/)
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.