Service oriented infrastructure (Part 2)


John Wittgreffe, Paul Warren

Welcome to Part 2 of this two part edition of the BT Technology Journal, dedicated to the theme of the Service Oriented Infrastructure (SOI).

The concept of SOI has been developed, by BT and other key players in the information and communication technology (ICT) arena, in response to the needs of customers. More and more of our enterprise customers are looking to the delivery of services as the way to achieve flexibility and economy. They want on-demand access to business functionality without the concomitant overhead and maintenance costs of their own internal ICT operations. The rise of Software-as-a-Service (SaaS), offering functionality ranging from customer relationship management to collaboration tools, is evidence of this. More recently we have seen computing infrastructure – that is, processing and storage – offered as a service across the internet. At the basic level, SOI is about providing ICT infrastructure as a service, but it is also about enabling the provision of SaaS and the creation of business processes from SaaS components.

In Part 1, we started by looking at the commercial drivers for SOI and discussing the fundamentals of SOI technology. We discussed the ICT infrastructure that underlies SOI, we described virtualisation technologies and we explored how these technologies can be orchestrated to meet application performance targets stipulated in Service Level Agreements (SLAs). We grounded all this with references to BT's own software and systems architecture.

Our story was also framed within the context of our three layer model for SOI, with the virtualised infrastructure at the bottom, a set of enabling services, including an Enterprise Service Bus (ESB), as a middle layer and the processes and end-user applications at the top. We also looked at some broader issues: at the meaning and implementation of SLAs in an SOI, for instance, and the emergence of a new discipline called service science that is studying the impact and consequences of the new service-oriented way of doing business. Finally, we considered some technologies which might be seen as alternatives to SOI, but are more likely to be complementary.

In Part 2, we build on this understanding to explain how SOI can be used to create a rich choice of services for the customer.

Infrastructure service providers, such as BT, are leaders in the provision of these services. However, a rich ecosystem of services will not be achieved by infrastructure service providers alone. To achieve a rich choice of service offerings, they must open up their underlying capabilities to third-party service providers. This leads to a model in which third-party service providers use infrastructure, delivered as a service from infrastructure providers, to create new services and cope flexibly with the varying demand for those services. In addition, to enable the implementation of SOI in the business world, the fundamental infrastructure must be complemented by features such as billing and security. Advanced technologies – real-time business intelligence and semantic technologies, for instance –have a role to play. The former can be used to automate the instantiation of resources and services in an SOI, to maintain the business-level SLAs which are important to the customer. The latter can be used to describe web services in a way that reduces the human effort involved in locating them and assembling them into composite services.

The goal is to enable the service oriented enterprise. This is illustrated in figure 1, which expands on the three-layer model we described in Part 1. At the bottom is the SOI itself. Here processing, storage and network resource are made available as virtualised resources across an Enterprise Service Bus (ESB). To the customer, this is all 'in the cloud', and therefore reachable anywhere. Above that, the ICT infrastructure is made available within the enterprise through its deployment of an SOA. This achieves flexibility by instantiating resources as required at run-time. Immediately above that, business applications are deployed through a Service-Oriented Architecture (SOA). At the next level up, those applications are combined to form business processes which extend across the enterprise boundary to create intimacy with customers.

In our first paper (page 13), Alan Ward and Chris Smith describe how service orientation is impacting BT's own product portfolio. They explain how the use of common capabilities enables the achievement of the seemingly conflicting objectives of reducing complexity for the service provider whilst improving choice for the service provider's customer. This is achieved through building blocks which are designed to have few and known interdependencies and which can be described in terms of their behaviour, performance and interfaces, not their implementation. Key factors in effective deployment of such an architecture are the selection and development of the correct building blocks and their subsequent management.

Figure 1. Supporting the service oriented enterprise

Figure 1. Supporting the service oriented enterprise

In the next paper (page 25), Simon Thompson discusses the value and characteristics of Software Development Kits (SDKs). He notes how opening up service development to users develops customer loyalty whilst also creating a valuable source of product innovation. The paper shows how a generic value chain framework can be used to characterise and compare SDKs. Using this framework, the author compares SDKs from Microsoft, Google, Apple, and BT. Finally, the paper discusses the various business strategies underlying the provision of SDKs and points to a future direction for BT.

For telcos themselves, the bottleneck in delivering new products most frequently occurs in the provisioning of service management support, such as billing, fault handling, and service set up. This is caused by a lack of automation, or at least a limited level of automation, in the provisioning systems. In response to this, the paper by Gary Bruce et al (page 35) describes the activities of the Product and Service Assembly (PSA) team, working under the auspices of the TeleManagement Forum (TMF). The paper notes that whilst there has been considerable development of Service Creation Environment (SCE) tools, there has been comparatively little work to develop technology to assemble and orchestrate the service management capabilities associated with a new product. It is this deficit which the PSA team is addressing. Central to their approach is the development of a reference architecture which decouples the detailed product knowledge from the monolithic support systems. The inspiration for this approach comes from practices within manufacturing sectors such as automotive and aerospace.

Rebecca Copeland (page 51) is concerned with the problems of rapid service creation in an IP Multimedia System (IMS) environment which is based on the same philosophy as SOA and SOI. She describes the Open System Innovation Platform (OSIP), in use at BT's laboratories. This platform highlights the importance of reusable and shared components with standard interfaces for rapid service introduction. The OSIP is intended for the exploration of novel multimedia services, utilising IMS applications together with existing BT capabilities. Through the OSIP service broker many web based services can be blended with telecommunications, providing a unified user experience.

Once created, a product needs to be managed through its lifecycle. The paper by Nektarios Georgalas and his colleagues (page 65) describes the introduction of the discipline of Product Lifecycle Management (PLM), already used in many high technology industries, into telecommunications. The paper discusses both intra- and inter-enterprise product lifecycles and describes a generic lifecycle pattern. Service Delivery Platforms (SDPs) allow the integration of networks, ICT and applications. The authors describe how the concept of the lifecycle pattern can be applied to such SDPs.

The next paper (page 87), by Paul Deans and Richard Wiseman, describes a demonstrator to illustrate how two real-life scenarios can be addressed by an SOI. Through these scenarios, the demonstrator illustrates how an SOI creates flexibility, visibility and control in the delivery of ICT services. This is achieved in particular through the use of a decision support system that orchestrates whatever corrective action is required and a dashboard that provides managers with visibility of system performance.

The flexibility of the SOI approach, and the resource sharing which it enables, creates a need for more flexible security processes and policies. Theo Dimitrakos, David Brossard and Pierre de Leusse (page 105) present security solutions developed at BT that address the challenges of security and dependability in an SOI. The paper particularly covers the required secure messaging and application gateways, federated identity management capability and service-level access management capability.

The range of on-demand services possible through SOI results in the need to provide accounting mechanisms for on-demand services. Whilst the consumer world is more used to on demand charging, in the business world ICT services are traditionally operated against fixed contracts with periodic (often monthly) resolution. James MacDonald (page 127) examines the key problems to be overcome, explains what features are required, and describes the prototype systems now being developed in BT. The goal is to achieve a rich variety of charging models for services including subscription, with strict end-to-end SLAs backed by real commitments to refund customers whenever agreed service parameters are not met.

The previous edition discussed the use of end-to-end application level SLAs in an SOI. This concept is taken further by John Shiangoli and Martin Spott (page 141) who show how business intelligence techniques can be used to manage SOI resources against the key performance indicators of business processes or strategic scorecards. In this approach, live data is used to learn relationships between the performance of strategic objectives and the underlying ICT infrastructure. Knowledge of these relationships is then used to manage that infrastructure to achieve those strategic objectives within constraints such as cost and customer experience.

In the final paper, John Davies et al (page 153) describe a world in which billions of parties are consuming services across the web. To achieve this vision, four technologies are needed:

  • service-orientation to support the development of complex services based on distributed and reusable components;
  • Web2.0 technology as a means to structure human-machine cooperation;
  • semantic technology for meaningful service discovery; and
  • web technology to provide the underlying infrastructure for the integration of services at world wide scale.

We are sure you will enjoy these papers. Together with those in the previous edition of our Journal, they provide a comprehensive overview of SOI, where it is now and where it will be in the future. Through these papers, we have seen how BT is in the forefront of developing the concepts of SOI and the associated tools. We look forward to further exciting developments and to BT's continuing role in helping to create them.

Paul WarrenPaul Warren works in BT's centre for information and security systems research where his interests centre on knowledge management, semantic technologies and the service oriented infrastructure. He is currently Project Director of ACTIVE (http://www.active-project.eu), a European project in the area of collaborative knowledge management. At an earlier stage in his career, Paul worked on technology strategy and technology foresight, investigating areas as diverse as e-business and novel forms of computing. His previous experience also included a secondment to the Confederation of British Industry to study government support for technology in industry. Paul has published widely on knowledge management, semantic technologies and technology foresight, and has collaborated in editing two books on topics related to ICT. He holds a degree in theoretical physics from Cambridge University and an MSc in electronics from Southampton University.

John WittgreffeJohn Wittgreffe works in BT's centre for information and security systems research. He has 20 years' experience in the industry, starting with optical networks in the late 1980s before moving to focus on migration planning for BT's network and service management systems. With the rise of e-commerce in the late 90s, John led a range of innovative e-business developments for corporate customers, including network-centric applications such as shopping, procurement, online voting, online aggregation, wireless applications and real-time solutions for IP-VPN management. He later combined his experience of internal operational support systems and contracts to lead capability strategy development for BT's global solutions business. John regularly speaks at industry conferences and consults with corporate customers regarding ICT futures. He currently runs R&D programmes in the areas of service oriented infrastructure and ICT management systems that aim to broaden and improve services for BT's enterprise grade customers. John holds a degree in physics from the University of York.