Consolidating to move ahead

In the 90s large companies decentralised and customised their ERP systems. Times have changed. Now it's time to work on the task of integrating many systems into one.

ERP

In the 90s large companies decentralised and customised their ERP systems. Times have changed. Now it’s time to work on the task of integrating many systems into one.

ERP consultants say that this type of “systems rollup” will be the predominant type of ERP project at large companies over the next five years. The aim is to have one instance of the enterprise software.

Analysts say one way to look at major ERP projects is as if they are railroad tracks. Two rails to follow; a technical one and a cultural one. Often the technical rail – the code, interfaces and documentation -- is easier to follow than the cultural one. Especially if various systems are spread across independent business units. It can be difficult to make decisions for the whole company. Instead companies may end up building consensus rather than making decisions which, according to project management gurus like Patrick Lenciona, is bad. It means you're trying to placate everyone and leads to mediocrity.

Combining separate ERP systems can mean that separate businesses will have to share processes and systems while accepting that they are accountable to the organisation as a whole. The payback for the organisation that can achieve this is that they get to decommission a range of separate systems and the costs that go with them. In the US some companies have realised operational savings on the 30% to 50% range by reducing the waste that comes from doing the same thing five different ways in five different business units. Analysts suggest that returns can be had in less than two years.

AMR Research analyst Jim Shephard says customer requirements such as single invoicing are driving companies to centralise their financials at least. They’re also realising a single view gives opportunities to cross-sell and upsell customers between business units.

Organisations with decentralised and customised ERP systems can do nothing, consolidate by division with governance or roll up into a single instance. Each is riskier and more expensive than the last, but each also carries exponentially greater benefits.

Building a business case is a long process – think one year rather than one month.

Vertical supply chain apps

Lessons learned from past supply-chain deployments helped enterprises realise that the deep industry knowledge and scalability offered by specialised systems can be key to streamlining supply chains. In response, best-of-breed vendors are building supply chain offerings tailored for specific vertical markets, such as consumer-packaged goods and processing manufacturing. Other companies are turning to their ERP vendors for supply-chain modules that snap on to existing accounting and HR systems, and enterprise application companies are mounting their own efforts to grab vertical business by touting the easy integration of their products to existing back-end systems.

Issues surrounding the integration of vertical solutions are often one of the deciding factors for companies weighing a specialised vertical supply-chain solution against a more horizontal solution, these actively marketed by ERP vendors such as SAP and Oracle that are designed to seamlessly link with their own ERP solutions at the back-end.

The rise of point solutions

According to CIO magazine in the US there is a groundswell of organisations seeking out specific point applications and backing away from ERP suites. This is a major shift from two years ago when everyone was investing in ERP and CRM packages. Numerous reasons are behind this sea change. Point applications cost less than enterprise suites and often offer deeper functionality. They are easier to implement and upgrade.

Of course, ERP suite vendors argue that integration and the hassle of managing multiple vendors makes point solutions more expensive over time. But with web services the theory (at least) is that integration will become easier and easier, so the main reason for buying a suite – ease of integration between functions – will cease to such a big motivator.

Compared with the traditional method of connecting point solutions via an interface, which must be done from scratch each time a new application is introduced, or implementing a big suite of applications that requires the IT department to rip out existing applications, web services can be a less complex and costly proposition.

Web services

Web services enable organisations to complete integration tasks cheaply and efficiently without the overhead that previously only large companies could afford, and allow data to be more easily shared between enterprise applications. One of the benefits people are seeing in web services technology is that it is an overlay on existing technology and doesn’t necessitate replacing what they already have. Last year major vendors started embracing web services with a vengeance and most new enterprise software will start shipping with web services interfaces. Already there is a proliferation of portal products that can display information from other applications via XML.

Every major vendor of CRM, ERP and business intelligence is either integrating or has plans to integrate web services into their systems.

Web services standards are still evolving and security remains an issue, but organisations may be willing to make that trade-off for increased flexibility.

Gartner research director Ray Wagner says so far, the major players and competitors in the web services market (BEA, Borland, Microsoft, IBM, Sun) have been willing to cooperate on developing security standards. They know the technology won’t gain broad acceptance without adequate mechanisms to protect privacy, confidentiality and the integrity of transactions.

However, he expects that as the industry’s attention turns to identity and proposed solutions such as Liberty Alliance and Microsoft Passport, as well as emergence of more complex extensions to the Web Services Security (WS-Security) specification, the standards battles will heat up significantly.

Co-operation between the vendors is already fraying and there are concerns that there might be fragmentation in certain web services standards such as through different implementations for Liberty Alliance and MS Passport supporters.

Standards bodies have been working together to create protocols and architecture guides for managing web services. Work is under way to align the architecture work of the World Wide Web Consortium (W3C), the Organisation for the Advancement of Structured Information Standards (OASIS) and the Common Information Model developed by the Distributed Management Task Force (DMTF).

OASIS plans to have a concrete protocol by the middle of the year, while the architecture guidelines from the W3C’s Management Task Force are expected to be ready about the same time. However, most vendors and analysts are saying not to wait around for standards. Get stuck in now by starting off small.

Last year industry players published a set of six advanced web services specifications that make it easier to apply business policy and implement security for a wider range of applications.

The new specs cover trust relationships, exchange of multiple messages, security policies, communication of requirements, and capabilities by senders and receivers of web services, attachment of a requirement and capability statements to web services, and general policies affiliated with a service.

They are now under public review and comment after which they will be revised and submitted to a standards body.

One of the proposed specs, proposed by Sun, along with BEA, Intalio and a number of other supporters, is a draft web choreography specification, called the Web Service Choreography Interface (WSCI), to the W3C. However, BEA, IBM and Microsoft published an alternative, the Business Process Execution Language for Web Services (BPEL4WS). Now Sun, Microsoft and IBM are sparring over which specification will become the standard. On the outside are vendors such as BEA and Oracle that are urging a convergence of the two specifications.

Users are interested in convergence because it would give them a greater range of tool choices and, more importantly, because the alternative could result in a nasty standards war. Industry commentators say "choreography" -- message handling -- standards are becoming increasingly important as company and customer business processes get more intertwined.

The main difference between WSCI and BPEL4WS is one of scope. WSCI is about web services choreography, while BPEL4WS includes choreography and orchestration specifications. This means that while both technologies cover the flow of messages between applications at a high level, BPEL4WS also talks about specifics, such as where to store incoming messages or what specific container to use as the body of the message.

Microsoft engineer and SOAP author Don Box, however, told developers at the XML Web Services One conference in the US last month that the software industry has become so fixated on new specifications that it has lost sight of the fundamental goal: using XML to link software applications together. While some new specs that have been proposed are important and useful, others are too complex

The rise of unauthorised web services

Gartner’s Wagner believes developers will rush to web services but they could often implement without giving enough consideration to security or efficiency. He says developers could start using them within organisations without the knowledge of CIOs and IT managers and predicts that by the second half of next year 40% of Global 2000 enterprises will have unauthorised, undocumented and unmonitored web services connections that extend beyond their perimeters. He says the sooner enterprises move to understand web services technology and establish monitoring and control policies for its use within and across their perimeters, the better off and less vulnerable they will be.

Analysts believe that complex, high-value web services deployments over the next 12 months will be rare. For a start complex, multiple-partner web services will require a large commitment of resources to security mechanisms, including public key infrastructure (PKI), which has hardly taken off at all.

Wagner says if you are considering a web services security platform make sure it can fully leverage standard SSL (secure sockets layer) server certificates. Not only for channel encryption, but also for two-way authentication and digital signing of transactions.

Enterprise application integration

When it comes to EAI New Zealand is still a fledgling market, according to consultant Aaron Kumove. EAI solutions are complex and the main vendors such as IBM, Microsoft, See Beyond, BEA, WebMethods and Tibco are still educating the market. There is a learning curve to overcome.

EAI is the integration of applications and data to enable sharing of data across applications without significant application change. People often ask, why not just use web services to do that?

One answer is that web services are all very well for connecting systems, but if you end up with a whole lot of web services going here, there and everywhere you’re inevitably heading for strife. Piecemeal point-to-point integration can be financially appealing but can result in an unwieldy mass of connections and interfaces that are difficult to manage and maintain.

All the major EAI vendors are adding support for XML and web services to their products. EAI software provides a way of orchestrating those services and managing the whole integration process. It can be seen as an evolution of middleware and adds efficiency to the integration process. Systems are connected to a broker which mediates all "conversation". The number of connections is reduced to the number of systems. They're all consistent and all speak in the same way to a broker.

EAI vendors recommended starting with small integration projects. Don’t try and integrate everything at once. Also requirements must be clear at the start.

One strategic approach is known as composite information system approach, which employs reusable interfaces. Legacy data is encapsulated into components, which can be CORBA, COM or .Net. Interfaces and components can be used reused. The client-side interfaces of a component don’t have to change, just because you change the back-end interface.

EAI isn’t easy. There is usually incompatible technology involved, the systems you are trying to integrate weren’t designed for interoperability, legacy systems are often poorly documented, data to be transferred into the new system is often incompatible, there are issues of security and people who “own” various applications may be suspicious of opening them up to other groups.

EAI software handles tasks such as data persistence, workflow design, orchestration, real-time tracking and monitoring of transactions message and queuing so that if a transaction can’t go through because one application is down or isn’t receiving as quickly as transactions aren’t being sent, they don’t drop out but are sent through when the receiving application is ready.

A broker or EAI solution usually presents immediate value but many organisations baulk from a cost perspective because to date they've been proprietary and the sticker prices have been high. Historically, EAI has been around for quite a long time in the form of messaging and then middleware, but its proprietary nature -- and high cost -- is being eaten away. Web services seem to be the key to this by standardising the functional stack, and as they do they are commoditising messaging. Until now companies have struggled to make the investment worthwhile on a single project. Analysts say the key to building a business case for EAI is to know the cost of integrating any application that you're developing and the cost of managing and maintaining it. Add these together and you should have a basis for comparing point-to-point development cost with an EAI off-the-shelf solution. In some cases you'll find EAI is cheaper on a single project and in some cases it isn't.

Where it's not justifiable on a single project you will get money back on the re-usability front, analysts say. The real value is on the second or third project. Incremental costs drop dramatically. IT analysts such as Gartner Group and Forrester suggest that up to 40% of the cost of developing a new application is spent on integration.

Join the newsletter!

Error: Please check your email address.

Tags enterprise software

More about AMR ResearchBEABorland AustraliaDistributed Management Task ForceGartnerGartnerIBM AustraliaLiberty AllianceMicrosoftOracleSAP AustraliaSee BeyondTibcoW3CWorld Wide Web Consortium

Show Comments
[]