Michael Hannaford

Stability in a Network of MONADS-PC Computers

The MONADS-PC computer system implements an architecture supporting a very large persistent store based on a uniform virtual memory. We have previously shown how this virtual memory scheme can be extended to encompass a local area network of MONADS-PC computers. In this paper we examine the question of the integrity of the store in such a network. A modification to the MONADS architecture to implement stability is reviewed and extended to guarantee stability of a network-wide persistent store. The stability scheme allows for temporary interruption to the physical network without affecting the validity of exported pages owned by a node.

Deadlock Detection Views of Distributed Database

Deadlock detection is very difficult in a distributed database system because no controller has complete and current information about the system and data dependencies. The deadlock problem is intrinsic to a distributed database system which employs locking as its concurrency control algorithm. This paper attempts a comprehensive study of deadlock detection in distributed database systems. Afterwards, a deadlock detection algorithm is presented. The algorithm is based on creating Linear Transaction Structure (LTS), Distributed Transaction Structure (DTS), finding local and global cycle, deciding priority Id of the transaction and local-global abortion. The proposed algorithm does not detect any false deadlock or exclude any really existing deadlocks. In this technique global deadlock is not dependent on the local deadlock.

Expanding the Cloud: A Component-Based Architecture to Application Deployment on the Internet

Cloud Computing allows us to abstract distributed, elastic IT resources behind an interface that promotes scalability and dynamic resource allocation. The boundary of this cloud sits outside the application and the hardware that hosts it. For the end user, a web application deployed on a cloud is presented no differently to a web application deployed on a stand-alone web server. This model works well for web applications but fails to cater for distributed applications containing components that execute both locally for the user and remotely using non-local resources. This research proposes extending the concept of the cloud to encompass not only server-farm resources but all resources accessible by the user. This brings the resources of the home PC and personal mobile devices into the cloud and promotes the deployment of highly-distributed component based applications with fat user interfaces. This promotes the use of the Internet itself as a platform. We compare this to the standard Web 2.0 approach and show the benefits that deploying fat-client component based systems provide over classic web applications. We also describe the benefits that expanding the cloud provides to component migration and resources utilisation.

Support for Object Oriented Transactions in Timor

An important aim in the design of the Timor programming language is to provide programmers with features which enable them to build complex systems from components which can be developed in isolation from each other (i.e. without knowledge of each others existence). The database transaction concept serves as an interesting test case for this objective, since it is a general concept which can be applied to many different applications. The paper discusses those features of Timor which allow this objective to be achieved.

Optimization of Detected Deadlock Views of Distributed Database

Deadlock is one of the most serious problems in multitasking concurrent programming systems. The deadlock problem becomes further complicated when the underlying system is distributed and when tasks have timing constraints. Deadlock detection and optimization is very difficult in a distributed database system because no controller has complete and current information about the system and data dependencies. The deadlock problem is intrinsic to a distributed database system which employs locking as its concurrency control algorithm. In this paper, an optimization technique for the detected deadlock is presented which minimizes the abortion of the selected victim transactions. The optimization technique is concerned with the detection of the transactions which are the basis for the most of the deadlock cycles (either local or global) in the system. The presented technique aborts the transaction’s requests which are really to blame for the formation of many deadlock cycles. Also the presented deadlock detection algorithm does not detect any false deadlock or exclude any really existing deadlocks. In this technique global deadlock is not dependent on the local deadlock system

Integrated Standard Environment for the Teaching and Learning of Operating Systems Algorithms Using Visualizations

Algorithm visualization is a technique used to help in the teaching and learning of algorithms in computer science. Despite the potential benefits of this technique, it has not been adopted by educators as expected due to the difficulties of integrating the algorithm visualizations into the learning environment and the lack of interactivity features. A course on operating systems constitutes one of the main courses in any undergraduate computer science program. Operating systems algorithms have not received enough support by the available algorithm visualization systems. Also, most of the available visualization systems developed to help in the teaching and learning of operating systems algorithms are standalone applications that do not provide important interactive features, such as the self-assessment of students’ knowledge. This paper presents a work in progress to design an interactive algorithm visualization system that can help to teach and learn operating systems algorithms. This system also provides self-assessment exercises to allow student to simulate how algorithms work and receive instant feedback from the system. This work adopts a new method to integrate algorithm visualizations into Learning Management Systems based on an e-learning standard called, the Sharable Content Object Reference Model(SCORM), which provides a set of specifications to design, package and distribute the learning content and standardize the communication between the learning content and the underlying Learning Management System. By using this standard, learning content can be integrated and reused by different Learning Management Systems.

A Distributed Content Storage Model for Web Applications

Vast quantities of information is now being stored online. Web applications currently rely on monolithic storage structures which place the sole responsibility of data storage, protection and maintenance on the web application provider. This research introduces the concept of a de-centralised approach for information storage online. Distributed storage techniques are used to address concerns with the classic monolithic approach while also addressing issues such as data ownership concerns for personal information. The research results in the presentation on an API that allows distributed storage of information with seamless integration of data into the traditional Web 2.0 model.

Querying Semistructured Data with Compression in Distributed Environments

As data management applications grow more complex, they are likely to need efficient distributed query processing. In Distributed Database Systems complete replication consists of maintaining complete copies of the database at each site; this has advantages such as highest locality of reference, highest reliability, availability, and is best for reading. The most promising and dominant data format for data processing and representing on the Internet is the semistructured data form termed XML. XML data has no fixed schema; it evolved and is self describing which results in management difficulties compared to, for example relational data. It is therefore a major challenge for the database community to design query languages and storage methods that can retrieve semistructured data. In this paper, we present a storing and querying scheme for semistructured data views of relational form in distributed environments. The proposed technique stores path dictionary, word dictionary, attribute dictionary, and the complete compressed replication of semistructured data in each distributed site of the DDBS. The presented technique provides query performance improvement due to the compression of semistructured data.

Isolation and Web Services Transactions

The traditional ACID properties for transactions are typically ignored in the Web services environment to ensure an acceptable level of service. However, this common lack of isolation can cause difficulties. We look at ways to reduce these problems but still maintain an acceptable level of service.

DRXQP: A Dynamic Relational XML Query Processor

The most promising and dominant data format for data processing and representation on the Internet is the semistructured data form termed XML. XML data has no fixed schema; it evolved and is self describing which results in management difficulties compared to, for example, relational data. It is therefore a major challenge for the database community to design query languages and storage methods that can retrieve semistructured data efficiently. In this paper, we present a querying scheme for semistructured data views of relational form. The proposed technique stores element-paths, attributes, contents of the element paths and attributes, and XML processing instructions in a dynamic relational structure termed as Multi-XML-Data-Structure (MXDS). We also analyse relational query processing time and XML query processing time to compare the query performance. The presented technique supports different types of XML query.