Integration and information go hand in hand. More than ever before, customers are demanding information; about their accounts, their balances, their recent purchases and deposits, their bills, their products - from every branch of your company. They want it over the phone, over the Internet or with the help of EDI or XML, and they want it now. Getting that information means accessing data from various systems, and on the CIO's to-do list that means integration. And so CIOs set out to integrate with all due haste; they use Cobol and legacy access tools, they sprinkle in screen scrapers and middleware, and they get the job done.
Unfortunately, cobbling systems together isn't enough. The question for CIOs to ask themselves isn't, "Are we integrating our systems?" That answer is yes for every company that hasn't yet been Delled or Amazoned out of existence.
The right question now is: "Can we get information from anywhere in our company to anywhere in our value chain?" The answer to that question is mostly a resounding no, and the vision for getting there - for becoming a truly integrated enterprise - is the real responsibility that CIOs must take on before writing another API or buying another Web/host integration package.
The vision for your integrated enterprise has to incorporate your company's infrastructure and logical connectivity to move data from application to application at speeds and volumes that will eventually dwarf today's requirements. Further, that vision has to make handling future integration needs easier, whether they arise from adding new e-commerce applications or from unforeseen mergers and acquisitions.
Any data, from anywhere to anywhere in the value chain, at any time: it's a tall order, but customer demands and competition from the dotcom world will push you to it. If you don't deliver, customers will eventually look elsewhere.
The integrated enterprise: delusional fantasy? Shimmering mirage in the blistering heat of the Internet economy? Not on your life. True, the standards aren't all settled yet for interbusiness communication, nor are the technologies mature. And organisational inertia is also an obstacle. "Large companies are structurally built to prevent [this degree of] integration," says Marc Cecere, a vice president at Giga Group (US). But it is clearly the direction in which today's businesses must go. The current practice of business is increasingly networked and interconnected. The benefits are numerous, and at the same time the retribution for those who fail to move with the current will likely be swift.
The first benefit of the integrated enterprise is the ability to deliver consolidated information to customers. Additional ROI comes in the form of lower costs of integration. "To me, the value proposition of a well-thought-out architecture is so compelling," says Lloyd Taylor, CIO of agricultural and industrial giant Cargill in Minneapolis. Taylor's company built its own message-based integration engine several years ago, and as Cargill developed interfaces to this central transport mechanism for each application, those interfaces went into a central library. "We could reuse these APIs, so the costs for integrating additional applications [throughout other Cargill units] started going down incrementally," says Taylor. As a result, the engine cut Cargill's integration costs from 50 per cent to 85 per cent on subsequent projects, he says.
But the potential benefits run deeper than cutting software costs. When connections to business partners are built into the integration plan, a higher order of efficiencies can be reached. Starting simply with electronic procurement, companies can move to disciplines such as collaborative demand forecasting and product design and development. Intertwining systems with suppliers and business customers allows companies to make dramatic improvements in their business processes, says Ted Rybeck, chairman of US-based consultancy Benchmarking Partners. And those processes and technology links can lock in profitable relationships for a long time to come. "I would like to be ordained by my customer as the partner of choice, and the worst possible place to do that is at the point of purchase," says Stan Elbaum, a partner at Benchmarking Partners. "I want to become part of their decision-making process."
Elbaum gives an example from his days in IS at Cott Corporation, a beverage company based in Toronto, Canada. In order to compete on price with behemoths Coca-Cola and Pepsi for business at major retailers, Cott needed to dramatically reduce its cost base. It redesigned its business processes in collaboration with its third-party bottlers. The resulting processes were not the cheapest way of doing business for Cott, but they lowered costs so dramatically for the bottlers that the total cost of delivering Cott's goods to the retailers became attractive.
To accomplish the data exchanges needed for the new way of doing business, Cott signed a contract with SAP reseller Siemens, allowing the bottlers to also install SAP's software, which the bottlers couldn't have afforded on a standalone basis. So Cott's integration plan included standardised data exchanges with supply chain partners, and the result was a dramatically cheaper way of doing business together.
In case all these carrots don't sufficiently entice you to engage in a full-fledged integration effort, dotcoms have the stick handy. Online enterprises have venture capital to burn on building integrated systems from the ground up. Indeed, their entire business models are based on information flows to such an extent that they sometimes have the opposite problem: Access is too easy. "We've got the credit card stuff clamped down, but if anything, we have too much information access," says Andy Martin, CTO of Garden.com, based in Austin, Texas. "Adding more security controls and constraints is our issue," he adds. Electronic communication and free-flowing data allows dotcoms the ability to customise the e-commerce experience for each customer as well as to build an operating model with lower cost structures than their older brethren.
Logically enough, end-to-end integration will be easier to achieve, cheaper and more reliable if the interconnections are put in place with a view to the whole. The integrated enterprise is a complex puzzle with many pieces.
Infrastructure Storage, bandwidth and processing power. A robust and connectivity-oriented infrastructure is an absolute must, and the volume of data to be stored, transported and processed will continue to grow. What will be stored online, nearline and offline, and in what formats? What data will be mirrored? How will integrity be ensured? "We won't be able to manage our data volumes in two years the way we do today," says Bruce Sink, CIO of First Union Corporation in Charlotte, North Carolina. Sink already tends 14 terabytes (plus a mirrored copy) of core storage from the company's transactional systems, and plans to use an additional petabyte - that's 1,125,899,906,842,620 bytes - of storage to handle cheque-image processing needs over the next five years. In fact, First Union is working in conjunction with three vendors, one apparently being storage giant EMC, to develop a new data management model. Sink won't disclose the details because the bank regards this model as a competitive advantage.
As for how all this data gets transported, Sink does not see the Internet as the universal pipeline for integrating the value chain, especially with the recent spate of denial-of-service hacker attacks. "The transport mechanism is the biggest issue from a technical perspective," says Sink. "How do you move three terabytes of information from point A to point B in a secure manner, and then query it without inundating [users] with data they can't use? Those are questions we don't have answers for yet," he says.
Ambitious Applications Installation of an ERP system most emphatically does not create an integrated enterprise. It may be a huge building block, but it isn't the whole story. Other crucial integrated functionality includes supply chain management, robust shop floor systems, customer-facing systems and some form of CRM or a customer life-cycle data warehouse. If applications for each of these functions can be bought off-the-shelf, and particularly if they have ready-made APIs that work with each other and with a central ERP system, then some of the integration challenge is reduced. Unfortunately, even the ERP vendors' supply chain management solutions are notoriously difficult to integrate with their own software. Also, most of today's CRM packages are quite limited in their scope, so that particular application set bears great scrutiny.
Middleware A single vendor isn't going to shrink-wrap the integrated enterprise, so some glue between applications is necessary. There are many approaches to application interoperability (see "Middleware Demystified", page 56). No single solution fits every need, and users commonly face a trade-off between flexibility and speed. A coherent approach to middleware selections is a crucial element, with an emphasis on flexibility and standardisation where possible.
The Will to Share Corporate culture is also key. Many companies, especially bigger and older ones, still labour against the reluctance of their own workforce to share information access. Unless people unlock their information vaults, all the technology is in vain. Consistently communicating the business benefits and the overall vision to executives will eventually garner the support necessary to open those doors.
Getting There From Here
The integration journey begins with a single step: the vision needs to be clearly and repeatedly communicated. That is the most commonly overlooked step in the scramble to keep up with IT demands.
After that vision is in place, then - and only then - can you get on with more typical project management activities like prioritisation. "There's probably one thing you'd better get done immediately, or a dotcom or competitor will come in and take it away from you," says Cecere. Identify the risks, rewards and urgency associated with each particular integration project, and prioritise the projects on that basis.
To handle the actual integration efforts, Cecere says smart companies are creating program offices, akin to the groups that handled Y2K conversion efforts. Cargill is a $US46 billion example of this approach. In 1995, the company's biggest division (oilseed processing) undertook an ERP implementation but wanted best-of-breed software in each function. That meant integrating systems from different vendors - a notoriously strenuous task with such complicated and ambitious applications as ERP. The company needed something more robust than its existing, home-grown messaging integration infrastructure.
And so Cargill formed an integration "centre of excellence" headed by Linda Ergen, enterprise application integration (EAI) manager, to offer the same expertise to other business units (which are responsible for their own IT applications). The company got consulting help from Hewlett-Packard for building a business plan and value proposition of the EAI centre. Ergen heads a group of 13 full-time employees who consult with the business units on architectural issues and develop the necessary software interfaces.
Call it a centre of excellence, a program office or what you will - a group of this sort puts the right kind of premium and visibility on integration work and brings together the corporate knowledge of what's going on in various technologies as well as in various arms of a big company.
Giga Group's Cecere argues against working too hard to define the perfect end state of all integration efforts. "You can spend an enormous amount of time planning, trying to solve the whole problem," he says. Cecere likens overly holistic integration planning to corporate data modelling efforts of the 90s, when trying to establish ironclad, corporatewide definitions of terms like customer, order and bill proved largely unprofitable in the face of in-house politics and changing technologies and standards. And it's clear that nobody has time to waste.
Still, the benefits of a holistic and careful integration plan are too powerful to ignore. CIOs who do not take enough time to develop such a plan will see the pain currently associated with integration continue to grow with soaring data demand and application proliferation, since every new integration project will require starting from scratch again. Any data, from anywhere to anywhere in your value chain, at any time - it sounds extreme now, but when has the demand for data ever decreased? And for those with a vision in place, future efforts will start with a solid technical foundation already laid.