the European Centre for Medium Range Weather Forecasting (ECMWF) in Reading, UK, held a workshop on Seamless Computing. Attending the workshop were more than fifty preeminent researchers and developers of systems and applications that required the use of high-performance distributed systems.
It was clear from the talks and discussions that occurred at the workshop that emerging applications require the ability to exploit diverse, geographically distributed resources. The reasons for this may be, for example, the location of specialized computational resources, availability of high-performance databases, high-speed network inter-connect, archival storage devices and so on.
The problem was, and is, how do we integrate resources of widely varying capabilities connected by potentially unreliable networks and often located in different administrative domains? This problem is being tackled by a number of international groups which are attempting to create either meta or seamless computing environments. Metacomputing and seamless computing have the same basic goals that of integrating distributed and heterogeneous systems into one integrated computing environment. Both need to deal with factors such as resource management and scheduling, file sharing, security, interoperability and scalability.
It became apparent at the workshop that there was a need to look at extensible frameworks that could be used to define the interfaces and interoperability layers between the disparate services available within such a wide-area environment. A follow-up Birdsof-a-Feather meeting was held at the SC97 conference in Santa Clara, CA. In this meeting further ideas were discussed and the call-for-papers for this special issue was made. These early seamless computing activities helped initiate the successful and on-going collaborative effort to enable Desktop Access to Remote Resources-DATORR.
The aim of this special issue is to gather together a body of work on all aspects of seamless computing. The four papers published here cover services to manage federated HPC systems, frameworks for job submission across wide-area heterogeneous systems, and a transparent and easy to use security system, designed for use in a wide-area environment.
It is evident that the emerging Java technologies and the Internet are transforming our view of how widely distributed systems should and could be used. It is to be hoped that the experiences of current day developers will ensure that our access and usage of wide-area distributed systems in the future is seamless, transparent and easy-to-use.