Singing from the same hymn sheet?

Enterprise application integration (EAI) is one of the burning issues facing companies as they strive to get the most out of their IT investments. Organisations, certainly those with multiple sites, have invariably built and inherited ranges of different systems that don’t work together. And with mergers and acquisitions the picture of complexity worsens – with businesses left wrestling with divergent technologies needing unification. Analyst Gartner Group reckons that integration will cost businesses world-wide billions of dollars over the next few years – and there’s a broad range of EAI solutions out there to soak up the inevitable budget, with virtually every serious software supplier having some kind of story in this regard. But while the integration challenge for many looks hard, a serious dose of pragmatism, favouring point-to-point integration projects, often makes relatively simple sense. According to Tim Jennings, research director at analyst Butler Group, since the ultimate goal is data visibility in whatever form and from whatever applications serve the key strategic and tactical needs of the company, big bang streamlining of the whole application base is rarely tenable or even necessary. “Information visibility is key to manufacturing as systems generate a glut of data that is often not in the right place or format to support business decision making,” he observes. “But companies often fail with [these projects] because they lack a strategic view first. The key is to have developed a framework, which is the premise of current EAI solutions anyway.” Surprisingly, he says manufacturing in general seems to be shying away somewhat from some of the EAI options apparently gaining popularity in other sectors. For example, servers sold by the likes of BEA Systems and IBM seem less favoured than using packaged ‘e’ systems layered on top of existing systems. But this view of EAI is supported by, for one, Nigel Bridges, director of information services at manufacturer Tyco Healthcare Europe. He insists that unless you first examine the business strategy there will be little or no reward from integration projects. Bridges explains that Tyco brought in consultancy PWC late in 1998 to have a good look at the business, highlight inefficiencies and suggest how centralisation could improve operations and how it might be achieved. “After two or three months of PWC analysis we saw a coherent business and systems strategy emerge. What we wanted was software that could run the company centrally, while serving local country needs and making us more tax effective,” says Bridges. “We also go through a high number of mergers and acquisitions [so] our prerequisite was not to have a 10 year project.” While other options were considered, Tyco went for a tactical solution with web-based software from Geac (formerly JBA Software) which delivered what Bridges calls a focused “enterprise supply chain”. Orders could be keyed in and relayed to existing local systems at the distribution centre best suited to pick and despatch product to local companies, with invoices triggered automatically. Says Bridges: “The only points of human intervention are the order entry and the picking and despatch of products.” It was a good, pragmatic solution – it took only six weeks to go live and has been acknowledged within Tyco as a huge success, with good attributable cost savings. In parallel to the Geac implementation, Tyco is also involved in Global Healthcare Exchange (GHX), an XML-driven private web marketplace for business-to-business (B2B) trading between large conglomerates in the sector. Here again, the GHX environment effectively serves another purpose hitherto performed by specific EAI software, namely providing the connecting ‘glue’ between multiple disparate trading systems – in this case, outside the organisation. Incidentally, where Tyco is a well established company with considerable complexity – the stamping ground of EAI software and service vendors – start-up firms facing the alternatives of buying ‘best of breed’ applications plus an off-the-shelf integration software framework, versus packaged solutions almost universally now go for the latter. Space4, a recent start up within the Westbury group, established in 2000 to make components for the house building industry, is a classic example. The firm wanted an integrated business system to handle manufacturing and financial management and connect to its e-business portal and CAM (computer aided manufacturing) systems. It went for a single environment from Navision. “I could argue that in its price range Navision’s individual modules compare well against individual ‘best of breeds’,” says Ken Packwood, e-business director at Space4. “However, Space4 is designed to have a low IT overhead. Wherever possible, outsourcing, mainstream, easily maintained and customisable systems were evaluated. Integration leads to specialist development and support, adds complexity and increases risk. Navision delivered what we wanted: why look for a more difficult solution?” Most would agree that today there is little contest here. Manufacturing malaise? That said, where multiple systems already exist – and that’s invariably the reality with trading supply chains, and often internal business and manufacturing management systems – the fact remains that too many in our sector, at least in the UK, seem still to be sweeping the issue under the proverbial carpet. According to Bill Pugsley managing director of e-business software vendor Perwill, while cash constraints are often blamed, actually its ignorance, lethargy and perhaps fear of change that are the biggest inhibitors to development. This company provides message brokering and integration software solutions developed through its key role in the Autoclear project with automotive sector manufacturer Mayflower, Wolverhampton University and Birmingham and Solihull TEC with European funding. Pugsley observes that manufacturers are looking for lower cost integration solutions, but says he is continuously frustrated by their ‘if it ain’t broke’ fixation and apparent failure to understand and implement modern, often lower cost technology. He laments a sector that seems intent on making cuts in its ERP and supply chain projects without first reconsidering some of the underlying processes that could reshape its companies’ competitive edge. “Only a few manufacturing companies are getting funds and these are being cautious, going for low cost projects with quick ROI (return on investment),” he says. “Some have even walked away from the web and developed unsophisticated interfaces to existing systems. They might use a web link, but are in fact no more than direct file transfer mechanisms between computers without using expensive EDI,” he says. The notion of web-based ‘loosely coupled’ integration of applications is favoured at the Butler Group. Jennings sees protocols like SOAP, or simple object access protocol, as the first layer to consider – providing a lightweight interoperability option and the “loose coupling of two organisations at the business process layer”. Jennings also sees application server platforms and XML brokers as currently mature routes worth considering. In particular, XML delivered via adapters or connectors, represents a good low cost way to exchange messages efficiently between systems in standard format, whether they are pre-built or customised, he notes. But the reality of adoption is as yet not great. Jennings reckons just 50% of companies are at the stage where internal point-to-point integration links are being achieved, while less than half that are embracing the business process issues, and less than 10% of companies are tackling full B2B web integration. He does not see some of the industry rumblings regarding the deterioration of XML as a standard, and believes there is better understanding in the market about the all important schemas, giving the markup language its self-describing quality. But it’s slow coming. Meanwhile, be warned: although the latest wave in integration is web services, it’s not the simple panacea that some claim. The real goal with web services is to deliver – in a purely business context – the kind of web application that has become prevalent on the public web. The aim is to allow people to delve deeply into back-end systems and provide self-service style access to information and transaction engines. Taking this route does not remove the need to reshape the IT architecture. Interestingly, an industry consortium has recently been formed to help ensure web services interoperability, with software vendors such as IBM, BEA and Microsoft in the vanguard. For his part, Jennings says he would prefer to see existing web standards organisations tackling the issues. But still a challenge All that said, it’s worth hearing what integration software solutions vendor SeeBeyond has to say. Ian Howells, vice president of marketing in EMEA for the firm, disagrees with Jennings and insists that, with a background of economic circumstances impacting companies and forcing consolidation, manufacturers are having to look for better management of everything from processes to product life cycles. That being the case, he believes that with the ERP (enterprise resource planning) task pretty well completed, the trend now is towards integrating existing applications. He also believes that supply chain streamlining is underway, again with EAI providing the linking technologies. Rest assured; either way there are limits. SeeBeyond too does not advocate total integration. Rather the goal should be solving EAI hotspots, such as those in supply chain planning where Howells reckons cost savings of between 30 and 60% are realistic. A real-time information network is the goal, he says, where processes are more visible and work faster, leading to lower levels of inventory, improvements in the accuracy of forecasting, ‘capable to promise’ throughout effectively synchronised supply chains and the rest. “This is a question of agility – in the UK companies often plan six months ahead, but what if the plan is wrong today?” However, if you have to go the bigger scale integrated environment route with a formal integration layer, as advocated by the purists, you need to take heed of vendors like WebMethods which offers end-to-end integration infrastructures. The target market for this firm is Global 2000 enterprises, and 40—50% of its customers are in manufacturing, so while the numbers aren’t great, the projects are large and hard experience has been gained. Laurie Mascott, vice president and general manager for EMEA at the company, says that before you do anything else, internal integration issues have to be faced. He mentions linking front and back office systems, legacy systems, databases, spreadsheets and management information systems – using data warehouses, the classic messaging systems, web hubs and the rest – as well as workflow and therefore business processes. All of it has to be considered, and a structured integration layer established early on. Mascott claims that his company’s approach is wholly focused on delivering web services, so it depends on achieving that fundamental integration layer which inevitably needs to involve standards, like XML, to bring about the requisite open architecture into which business processes can be plugged. He also advises manufacturers to keep their goals in mind, citing creating a seamless supply chain and building a platform for better relationships with customers and suppliers as the most common – by effectively extending ERP to become the supply chain lynchpin. For his own firm’s services, Mascott claims rapid ROI as 30 to 90 day projects are realistic, with six to nine months for integrating the entire backbone. Altogether though, apart from the largest manufacturing companies, it is clear that there is currently little stomach for large scale IT projects of any kind. We should hardly be surprised given the long march towards fully functional ERP, Y2K and the current economic climate. Yet much can be gained from small scale, focused, tactical integration projects at one end of the spectrum, and strategic web links at the other. If cash is short, spending it on what matters most is absolutely a good option.