Theory and practice of building reliable distributed applications

Abstract

Service providers are looking to computer vendors to provide low-cost, scalable faulttolerant solutions. The prime requirement is to minimise reliance on specialist equipment and techniques for delivering core services. Indeed, an ideal solution would make use of ‘standard’' middleware services (e.g., CORBA services for persistence, transactions etc.). Research results on distributed objects and software implemented fault-tolerance techniques hold the promise of providing such solutions. However, the task of constructing such solutions using general-purpose, low cost components, such as commodity UNIX servers, middleware services etc. is extremely challenging. The central problem is that any software implemented distributed fault-tolerance technique consumes resources (a combination of network bandwidth, processing power and disk storage) that otherwise would be available for normal use. Thus software implemented distributed fault-tolerance techniques must be applied with care. In this tutorial, we will cover basic principles of software implemented fault-tolerance techniques and discuss their application in the construction of reliable distributed applications. Emphasis will be placed on applications in the Internet/Web domain. Mark Little is a Distinguished Engineer/Architect, within HP Arjuna Labs., Newcastle upon Tyne,England, where he leads the Total-e-Transactions team. He joined HP via a series of company acquisitions: Bluestone Software, Arjuna Solutions, of which he was one of the founders. Before joining Arjuna Solutions he was for over 10 years a member of the Arjuna team within the University of Newcastle upon Tyne (where he continues to have a Visiting Fellowship). His research within the Arjuna team included replication and transactions support, which include the construction of an OTS/JTS compliant transaction processing system. Santosh Shrivastava was appointed a Professor of Computing Science, University of Newcastle upon Tyne in 1986; he leads the Distributed Systems Research Group. He received his Ph.D. in computing science from Cambridge in 1975. His research interests are in the areas of distributed systems, real-time systems, fault tolerance and application of transaction and workflow technologies to e-commerce. His group is well known as the developers of an innovative distributed transaction system, called Arjuna and a CORBA based dependable workflow system for the Internet. Together with his colleagues he set Proceedings of the Third International Symposium on Distributed-Objects and Applications (DOA’01) 0-7695-1300-X/01 $10.00 © 2001 IEEE up Arjuna Solutions Ltd in 1998 in Newcastle to productise Arjuna transaction and workflow technologies. The company is now part of Hewlett-Packard. Stuart Wheater is a Distinguished Engineer/Architect, within HP Arjuna Labs., Newcastle upon Tyne, England. He joined HP via a series of company acquisitions: Bluestone Software, Arjuna Solutions, of which he was one of the founders. Before joining Arjuna Solutions he was for over 10 years a member of the Arjuna team within the University of Newcastle upon Tyne (where he continues to have a Visiting Fellowship). His research within the Arjuna team included transactions and long-lived process support, which includes the construction of a CORBA based transactional workflow system. Proceedings of the Third International Symposium on Distributed-Objects and Applications (DOA’01) 0-7695-1300-X/01 $10.00 © 2001 IEEE