Paper pdf

Presentation slides

The Koala Component Model for Consumer Electronics Software

Abstract. A component-oriented approach is an ideal way to handle the diversity of software in consumer electronics. The Koala model, used for embedded software in TV sets, allows late binding of reusable components with no additional overhead.

Paper pdf

Presentation slides

ASpectJ2EE

Abstract. J2EE is a middleware architecture augmented with supporting tools for developing large scale client/server and ulti-tier applications. J2EE uses Enterprise JavaBeans as its component model. The realization of these components by a J2EE application server can be conceptually decomposed into distinct aspects such as persistence, transaction management, security, and load balancing. However, current servers do not employ aspect-oriented programming in their implementation. In this paper, we describe a new aspect language, AspectJ2EE, geared towards the generalized implementation of J2EE application servers, and applications within this framework. AspectJ2EE can be easily employed to extend the fixed set of services that these servers provide with new services such as logging and performance monitoring. Even tier-cutting concerns like encryption, data compression, and memoization can be added while avoiding the drags of cross-cutting and scattered code. AspectJ2EE is less general (and hence less complicated) than AspectJ, yet demonstrably powerful enough for the systematic development of large scale (and distributed) applications. The introduction of parameterized aspects makes aspects in AspectJ2EE more flexible and reusable than aspects in AspectJ. AspectJ2EE also generalizes the process of binding services to user applications in the application server into a novel deploy-time weaving of aspects. Deploytime weaving is superior to traditional weaving mechanisms, in that it preserves the object model, has a better management of aspect scope, and presents a more understandable and maintainable semantic model.

Paper pdf

Presentation slides

Resolving Feature Convolution in Middleware Systems

Abstract. Middleware provides simplicity and uniformity for the development of distributed applications. However, the modularity of the architecture of middleware is starting to disintegrate and to become complicated due to the interaction of too many orthogonal concerns imposed from a wide range of application requirements. This is not due to bad design but rather due to the limitations of the conventional architectural decomposition methodologies. We introduce the principles of horizontal decomposition (HD) which addresses this problem with a mixed-paradigm middleware architecture. HD provides guidance for the use of conventional decomposition methods to implement the core functionalities of middleware and the use of aspect orientation to address its orthogonal properties. Our evaluation of the horizontal decomposition principles focuses on refactoring major middleware functionalities into aspects in order to modularize and isolate them from the core architecture. New versions of the middleware platform can be created through combining the core and the flexible selection of middleware aspects such as IDL data types, the oneway invocation style, the dynamic messaging style, and additional character encoding schemes. As a result, the primary functionality of the middleware is supported with a much simpler architecture and enhanced performance. Moreover, customization and configuration of the middleware for a wide-range of requirements becomes possible.

Paper pdf

Presentation slides

Collaborative Reinforcement Learning of Autonomic Behaviour + DOA retrospective

Abstract. This paper introduces Collaborative Reinforcement Learning (CRL), a coordination model for solving system-wide optimisation problems in distributed systems where there is no support for global state. In CRL the autonomic properties of a distributed system emerge from the coordination of individual agents solving discrete optimisation problems using Reinforcement Learning. In the context of an ad hoc routing protocol, we show how system-wide optimisation in CRL can be used to establish and maintain autonomic properties for decentralised distributed systems.

Paper pdf

Presentation slides

DOA retrospective slides

Abstract. In this paper we describe the Pegasus system that can map complex workflows onto the Grid. Pegasus takes an abstract description of a workflow and finds the appropriate data and Grid resources to execute the workflow. Pegasus is being released as part of the GriPhyN Virtual Data Toolkit and has been used in a variety of applications ranging from astronomy, biology, gravitational-wave science, and high-energy physics.

Paper pdf

Presentation slides

Presentation slides

Abstract. The Grid is a dynamic environment in which resources can quickly go from idle to busy state depending on application operations. In such a scenario, resources can be used more effectively by introducing reservation and allocation. As a solution, this paper proposes STREGA: a software architecture that handles resource reservation and greatly simplifies the integration of applications with a Grid environment. In it, resources needed by applications are automatically detected, and operations such as resource reservation and allocation are accordingly transparently performed e.g. using Globus services. Within STREGA, some components are aimed at understanding the needs of application classes, other components dynamically re-adapt resource requests on the basis of the observed application behaviour. Additional components are proposed to support reservation when this is unavailable from the underlying system (i.e. Globus and the OS).

Presentation slides

Abstract. A large number of structured peer-to-peer overlay systems have been proposed recently. Due to their scalability, robustness and self-organizing nature, these systems provide a very promising platform for a range of large-scale, distributed applications. Due to the lack of a publicly available shared structured peer-to-peer infrastructure, current applications using such functionality need to implement and provide it themselves, resulting in significant inefficiencies and duplication of work. The main challenge in implementing such a shared infrastructure lies in the design of a universal API that meets the needs of all current and future applications. We believe it is impossible to statically define such a universal interface without either being highly inefficient or potentially hindering the deployment of future applications with new and unpredictable requirements. In this position paper, we propose a programmable and dynamically extensible structured peer-to-peer infrastructure that is highly flexible and allows easy and incremental deployment of new applications and services on top of it. We outline the architecture of such a system and discuss its benefits as well as challenges that need to be addressed.

Paper