Many Grid applications can benefit from reliable multicast (RM). One example is a simulation that generates Terabytes of data, which needs to be reliably transmitted to multiple visualization servers (and archiving machines), and from these to clients. There has been much research on scalable RM protocols, and the literature has many examples of protocols and analytical/simulation comparisons. Unlike previous work, this paper describes an experimental evaluation of a selected set of reliable multicast protocol implementations, in the context of Grid applications. The experiments are performed on UK’s high-speed academic network SuperJanet. Our results show that the use of these protocols greatly reduces the network overhead that is incurred in point-to-multipoint transfers using TCP. However, we found that the current implementation of the protocols fail to satisfy the high-throughput requirements of the Grid. 1.

Grid applications typically deal with huge amount of data and often the same data have to be transferred and processed on many resources. Nevertheless, the majority of existing middleware platforms for Grid computing do not provide suitable programming and communication models to make easy software development and to improve communication performances when a large set of receivers is involved. Some middlewares for wide area network computing, such as ProActive, provide the group abstraction to transparently deal with a number of similar receivers. We propose an extension of such a mechanism in order to improve its features for Grid environments. In particular, ProActive native groups have been extended both at programming and communication levels in order to support both different internal behaviors and high performance communication based on IP multicast. A case study shows the effectiveness of the new mechanism and its efficiency compared with the original one.

Permission to make digital or hard copies of portions of this work for personal or classroom use is granted provided that the copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise requires prior specific permission by the publisher mentioned above.

In its simplest form, multicast communication is the process of sending data packets from a source to multiple destinations in the same logical multicast group. IP multicast allows the efficient transport of data through wide-area networks, and its potentially great value for the Grid has been highlighted recently by a number of research groups. In this paper, we focus on the use of IP multicast in Grid applications, which require highthroughput reliable multicast. These include Grid-enabled computational steering and collaborative visualization applications, and wide-area distributed computing. We describe the results of our extensive evaluation studies of state-of-the-art reliablemulticast protocols, which were performed on the UK’s high-speed academic networks. Based on these studies, we examine the ability of current reliable multicast technology to meet the Grid’s requirements and discuss future directions.

Much work has been done attempting to scale multicast transmission to hundreds or thousands of receivers, but typical applications requiring bulk data transfer may involve just ten or twenty sites. We have been investigating how multicast can be exploited in such an environment and have developed a small-scale sender-initiated reliable multicast protocol called TCP-XM. In order to test and evaluate this new protocol we have used it to build a reliable multicast file transfer tool called mcp. In this paper we describe the mcp transfer tool, review the design of the TCP-XM protocol, and provide experimental results from tests based on both network emulation and live wide area network transfers between sites on the UK eScience network. 1

This article presents an object-oriented mechanism to achieve group communication in large scale grids. Group communication is a crucial feature for high-performance and grid computing. While previous work on collective communications imposed the use of dedicated interfaces, we propose a scheme where one can initiate group communications using the standard public methods of the class by instantiating objects through a special object factory. The object factory utilizes casting and introspection to construct a “parallel processing enhanced ” implementation of the object which matches the original class ’ interface. This mechanism is then extended in an evolution of the classical SPMD programming paradigm into the domain of clusters and grids named “Object-Oriented SPMD”. OOSPMD provides interprocess (inter-object) communications via transparent remote method invocations rather than custom interfaces. Such typed group communication constitutes a basis for improvement of component models allowing advanced composition of parallel building blocks. The typed group pattern leads to an interesting, uniform, and complete model for programming applications intended to be run on clusters and grids.