The Requirements Engineering Specialist Group
of the British Computer Society
:
About
Join the RESG
Mailing List
Events
RQ Newsletter
RE Sources
Search & Map
Home
Contact Us
Past RESG Event
Service-Centric RE
Speakers' slides available below.
Date
07 March 2007, 1.30pm
Venue
City University, London
Birley Lecture Room, entrance on Spencer St. Location 5 on this map.
Service-centric systems raise new challenges for how analysts discover and manage requirements. Some of these are similar to the challenges raised by component-based systems, such as the problem of matching required properties to component or service capabilities. However, services raise a set of new challenges. These are a consequence of factors that include:
- Service consumers purchase a service that executes remotely, not a component that executes locally;
- Service discovery is itself a computational activity that may be performed under the control of an analyst at design time, but may be performed automatically at run-time should (e.g.) a service failure necessitate re-binding;
- Current service specification and discovery protocols capture no semantics or quality information.
This afternoon event will explore a number of these issues using experience from large-scale service-based system development projects and current research.
Program
13.30
Registration
14.00
Introduction and welcome
Neil Maiden & Peter Sawyer
14.10
The Role of Service Level Agreements in Service Requirements Specification
Olivier Nano, European Microsoft Innovation Centre
Slides (pdf)
14.40
Discovering Web Services To Specify More Complete System Requirements
Konstantinos Zachos, Neil Maiden, Xiaohong Zhu, Rhydian Howells-Morris and Sara Jones, City University, London
Slides (ppt)
15.10
Tea/Coffee
15.30
Specifying Service Qualities
Glen Dobson and Pete Sawyer, Lancaster University
Slides (ppt)
16.00
Requirements for an Integration Broker for Heterogeneous Information Sources
David Budgen, Department of Computer Science, University of Durham
Slides (pdf)
Talk Abstracts
Discovering Web Services To Specify More Complete System Requirements
Konstantinos Zachos, Neil Maiden, Xiaohong Zhu, Rhydian Howells-Morris and Sara Jones
Centre for HCI Design, City University, London
Service-centric systems pose new challenges and opportunities for requirements processes and techniques. This presentation reports new techniques developed by the EU-funded SeCSE Integrated Project that enable service discovery during early requirements processes and exploit discovered services to enhance requirements specifications. It presents the algorithm for discovering services from requirements expressed using structured natural language, and demonstrates it using an automotive system example. The presentation also reports a first evaluation of the utility of the environment that implements this algorithm when improving the specification of requirements with retrieved services.
Specifying Service Qualities
Glen Dobson and Pete Sawyer
Computing Department, Lancaster University
In engineering Service-centric Systems the integrator requires the ability to compare the quality of competing services. It must therefore be possible to state quality requirements in a way that allows easy comparison with service specifications, descriptions, monitoring data, SLA (Service Level Agreement) offers, etc. The interactions which occur during this comparison process are increasingly facilitated by intelligent discovery, brokering and monitoring software. A machine-understandable QoS (Quality of Service) model is therefore highly desirable. "Understandable" in this sense implies more than the ability to match syntax.
Consider, for instance, that a requirement is stated for availability of at least 99.99%. If a service provider or monitor instead states a mean time to recover of 2 minutes, and a mean time between failures of 14 days then a simple syntactic comparison with the availability requirement will lead to the incorrect conclusion that the requirement is not satisfied by the service in question. Equally, a requirement stated for a mean time between failures of more than a week will also result in a false negative in this case, due to the use of different temporal units. As one examines the problem further, more and more important semantics of QoS emerge.
A knowledge-based approach to QoS requirements therefore becomes increasingly appealing, as it allows machines to "understand" the semantics which are neglected in existing approaches. In our talk we will discuss our formal QoS ontology which forms the basis of such a knowledge-based approach to QoS; the advantages which can be achieved through its use; and the role of an inference engine in providing many of these advantages.
Requirements for an Integration Broker for Heterogeneous Information Sources
David Budgen
Department of Computer Science, University of Durham
The IBHIS project (Integration Broker for Heterogeneous Information Sources) originally set out to investigate a problem that is becoming more commonplace: namely, how to obtain and integrate information (about some individual) that is obtained from a set of autonomous and independent organisations, stored in a multiplicity of formats and subject to externally-determined rules about access. The domain chosen to provide our ‘case study’ material was that of healthcare, where there are many instances that demonstrate the need to integrate patient information that may in turn be drawn from a range of different sources.
Our basic postulate was that this type of problem had features that were best met by a service-based architecture. Indeed, we could argue that IBHIS is an instance of a ‘classical’ services architectural pattern.
To give the requirements elicitation a more systematic basis we worked with healthcare specialists to create a set of six use cases that described very different forms of episode of healthcare. We then used these to help explore both the problem and the appropriate form of solution.
In my presentation I will briefly explore the process of eliciting the use cases, and then explain how these helped us explore what was required of our service architecture, and the form that this took. I will also briefly examine the ways in which we also sought to use service characteristics to ensure that our ‘broker model’ was suitably generic and reusable in other domains.
The Role of Service Level Agreements in Service Requirements Specification
Olivier Nano
European Microsoft Innovation Centre (EMIC)
Today, more and more applications are delivered through services.
Services Oriented Architecture (SOA) is very valuable for loosely
coupling different parts of a system. It adds agility to respond to
environment evolution, it helps to quickly provides new functionalities.
But in order to add business viability, it is important to be able to
describe the expected business
level objectives. A Service Level Agreement (SLA) is a contract that
links the service provider and the service consumer on the business
level objectives of the service. In this talk we'll cover the SLA life
cycle in a service centric system, we'll describe what are the
properties needed for an electronic SLA and we'll discuss the open
challenges to make SLA a first class citizen in service centric systems.
[Back]