Grid and P2P Paradigms for Parallel Processing

Grid computing emerged as a paradigm for high‐performance computing and massive parallel processing. Currently, Grid systems have become an important paradigm for efficiently solving large‐scale complex problems from many fields. On the other hand, P2P paradigm originated from file‐sharing, but each time more is being used for the development of large‐scale distributed platforms. Grid and P2P systems have thus followed different trajectories pushed by different motivations, needs and research communities. In fact, both paradigms are evolving in a way that each time they are sharing more common characteristics and are mutually benefiting from their best features. Among these characteristics, we could distinguish the cooperative model for solving complex problems by exploiting their large computing capacity. As such, Grid and P2P systems have achieved notable success, in particular, for e‐Science applications, a family of complex applications arising in science and engineering that need considerable computation power. Despite important advances in the design and use of Grid and P2P systems, they remain still difficult to implement and apply to real‐life problems. The main difficulties reside in the lack of easy‐to‐use middleware for Grid and P2P, in the complexities of setting up and in the tedious task of deploying real‐world Grid/P2P platforms as well as in experimental studies which are often complex and not easy to repeat. In this paper we survey and analyze the advances in communication libraries and middleware for both Grid and P2P systems as well as their limitations when used in real Grid and P2P infrastructures. We also bring examples of real‐life applications of massive data processing that can be efficiently handled through Grid and P2P approaches. Copyright © 2010 John Wiley & Sons, Ltd.

Heterogeneous Network-Based Concurrent Computing Systems

Peer-to-Peer resource discovery in Grids: Models and systems

Grid and P2P systems are different instances of distributed systems: while Grids are often motivated by intensive computations and required access to huge amounts of distributed data in a secure and trusted environment with quality of service guarantees, P2P systems are driven by large-scale dynamic collaborations in highly untrusted and unreliable environments. Since 2001, the Global and Peer to Peer Computing Workshop has been held annually along with the IEEE Cluster Computing and the Grid (CCGRID) Conference to capture the fuzzy border between these areas and blur it even more, showing a much welcome exchange and enrichment of research ideas. The workshop popularity has increased continuously, making it one of the most successful workshops held at the CCGRID conference. This special issue includes extensions of the best papers of the 2005 workshop issue along with papers submitted directly to the journal. We received a total of 27 submissions and selected 7 papers for their high level scientific merits. One of the often cited differences between Grids and P2P systems is scale. While grids were traditionally developed around small and close professional communities, P2P systems were from the beginning intended to allow millions of anonymous users to share resources. However, the last years demonstrated an increased interest in scalable and large-scale grids. The articles in this special issue address the problem of larger grids from different directions, contributing ideas and software that may eventually take grids to truly global scale. The first direction is unrestricted participation: grids are often collaborations specialized on a set of predefined projects with limited access to participation. OurGrid, the system presented in “Labs of the World, Unite!!!”, is a free-to-join, already deployed peer-topeer grid. Anyone can join it to gain access to large amounts of computational power collected from the idle resources of all participants. Secure access to resources relies on an incentive mechanism that encourages participation and sharing, and uses virtual machines for resource isolation. An important problem stressed in large-scale systems is trust: while trust can be assumed through out of band means in small, controlled communities, it needs to be explicitly built and maintain in larger, more open groups. Two articles in this issue attack the problem of trust: in “Applying a Trust Brokering System to Resource Matchmaking in Public-Resource Grids,” Azzedin, Maheswaran and Mitra propose a trust J Grid Computing (2006) 4: 223–224 DOI 10.1007/s10723-006-9053-5

Computational Grid and Peer to Peer computing system are interlapping with each other progressively. This paper brings forward a resource model for future large-scale non-hierarchy grid system. With this model we can represent heterogeneous resources sharing relationships. Based on the assumption of power-law degree distribution and result of Lada A. Adamic [20], we find that the unstructured locating algorithms used in P2P system do not suit this resource model. Finally we suggest that it is necessary to classify Grid systems, observe network topology and build classified resource models for different types of Grid system.

Computational grids are an important emerging paradigm for large-scale distributed computing. As grid systems become more wide-spread, techniques for efficiently exploiting the large amount of grid computing resources become increasingly indispensable. A key aspect in order to benefit from these resources is the scheduling of jobs to grid resources. Due to the complex nature of grid systems, the design of efficient grid schedulers becomes challenging since such schedulers have to be able to optimize many conflicting criteria in very short periods of time. This problem has been tackled in the literature by several different metaheuristics, and our main focus in this work is to develop a new highly competitive technique with respect to the existing ones. For that, we exploit the capabilities of cellular memetic algorithms (cMAs), a kind of memetic algorithm with structured population, for obtaining efficient batch schedulers for grid systems, and the obtained results will be compared versus the state of the art. A careful design of the cMA methods and operators for the problem yielded to an efficient and robust implementation. Our experimental study, based on a known static benchmark for the problem, shows that this heuristic approach is able to deliver very high quality planning of jobs to grid nodes and thus it can be used to design efficient dynamic schedulers for real grid systems. Such dynamic schedulers can be obtained by running the cMA-based scheduler in batch mode for a very short time to schedule jobs arriving in the system since the last activation of the cMA scheduler.

Are P2P systems and applications here to stay? Or are they a bright meteor whose destiny is to disappear soon? In this paper we try to give a positive answer to the first question, highlighting reasons why the P2P paradigm should become an integral part of computing and communication services and not only oddities for Cyber-geeks.

This work proposes a new parallel processing paradigm using a reconfigurable computing system. In proposed system, an application program is parallelized statically into threads. They are translated to hardware threads, and then, they are executed by a reconfigurable computing system directly. The hardware threads cooperate with a high-speed synchronization using tags on the reconfigurable device. To evaluate the effect of a parallel processing and the performance impact of synchronization, we have performed the preliminary experiments using some application programs on a real machine. The result shows that extracting parallelisms improves the performance of 1.37 times at average. The low-overhead communication and synchronization on the reconfigurable device can achieve good performance improvement on a parallel processing.

Recent advances in the power of parallel computers have made them attractive for solving large computational problems. Scalable parallel programs are particularly well suited to Massively Parallel Processing (MPP) machines since the number of computations can be increased to match the available number of processors. Performance tuning can be particularly difficult for these applications since it must often be performed with a smaller problem size than that targeted for eventual execution. This research develops a performance prediction methodology that addresses this problem through symbolic analysis of program source code. Algebraic manipulations can then be performed on the resulting analytical model to determine performance for scaled up applications on different hardware architectures. >

Most known methods for the determination of the structure of macromolecular complexes are limited or at least restricted at some point by their computational demands. Recent developments in information technology such as multicore, parallel and GPU processing can be used to overcome these limitations. In particular, graphics processing units (GPUs), which were originally developed for rendering real-time effects in computer games, are now ubiquitous and provide unprecedented computational power for scientific applications. Each parallel-processing paradigm alone can improve overall performance; the increased computational performance obtained by combining all paradigms, unleashing the full power of today's technology, makes certain applications feasible that were previously virtually impossible. In this article, state-of-the-art paradigms are introduced, the tools and infrastructure needed to apply these paradigms are presented and a state-of-the-art infrastructure and solution strategy for moving scientific applications to the next generation of computer hardware is outlined.

Special issue on Grid and P2P Systems

Data Management in Cloud, Grid and P2P Systems

Current representatives of Grid systems are Globus and Web Services, however, they have poor scalability and single point failure. It is a hot research topic to build a P2P and grid hybrid framework for resource management and task schedule. We propose Information Pool Based Grid Architecture (IPBGA), which is a real sense hybrid of P2P and grid instead of only introducing P2P methods into grid systems for resource management. The key of IPBGA, information pool protocol, is presented. In our information pool protocol, all of resources and tasks are abstracted into information, and resource requests for tasks and task appeals for resources are viewed as information services, then grid resource management and task schedule are treated as information matching. Therefore, our architecture is very adaptive to heterogeneous, dynamic, and distributed grid systems. We use tri-information centers for collecting information, which strengthens the robustness of our system. Simulation experiments show information pool protocol of IPBGA is more efficient in resource management and task schedule, and has less bandwidth and processing cost compared with other hybrid P2P systems.

Computer systems are characterized by an ever-growing complexity and a pronounced distributed nature. Since controlling highly distributed systems and managing the communication among them are far beyond the capabilities of a central entity, it is essential to develop new decentralized architectures. Such architectures, for example Grids, Clouds and P2P systems, are increasingly popular, but they need new types of algorithms to be efficiently managed. The recent paradigm of Organic computing (OC) concerns a large collection of intelligent (embedded) systems with potentially unlimited networks and spontaneous local interactions (Merkle et al. 2008; Muller-Schloer et al. 2011). The goal of OC is to develop new concepts and tools to manage the complexity of such systems, which must be designed to be adaptive, self-organized and manageable at the same time (Schmeck 2005). Adaptive OC systems must be capable of learning and adjusting their behaviour to dynamically changing environments. In addition, they should have self-x properties such as self-healing, selfdefining, self-configuring and self-optimizing. These properties are inspired by the mechanisms used by the autonomic nervous system to regulate the body without conscious input from the individual. Bio-inspired algorithms and techniques feature fault-tolerant and self-adaptive behaviours that help to boost the autonomic nature of distributed systems (Germain-Renaud and Rana 2009), and are proving effective to solve highly parallel and distributed problems. These techniques sometimes rely on the operations of agents, whose behaviour is inspired by biological systems, including ant colonies, bird flocks, honey bees, bacteria, and many more. In such systems, ‘‘swarm intelligence’’ emerges from the interaction of a large number of very simple agents. Bio-inspired algorithms have been successfully used as alternative and/or superior solutions to the problems that traditional algorithms cannot solve satisfactorily. In recognition of their achievements and potential, bio-inspired algorithms were named one of Scientific American magazine’s 10 ‘‘World Changing Ideas 2010’’. Bio-inspired algorithms and systems are applied to hard and large problems in a variety of areas. Some examples are optimization problems solved with genetic algorithms, routing strategies inspired by honey bee behaviour, ant-inspired resource discovery and data mining computations in Grid, Cloud and P2P frameworks, and so on. The papers included in this special issue show recent results in the area of bio-inspired distributed computing. The authors presented their preliminary works in the 3rd Workshop on Bio-Inspired and Self-* Algorithms for Distributed Systems, BADS 2011 (Folino et al. 2011), which was held together with the International Conference on Autonomic Computing (ICAC 2011) in Karlsruhe, Germany, in June 2011. The three papers were first selected among the papers presented in BADS 2011; then, they were extended and refined through two rounds of reviews. The paper ‘‘Protein structure prediction using distributed parallel particle swarm optimization’’, by Ivan Kondov, demonstrates the efficiency of the standard Particle swarm optimization (PSO) algorithm in the hard task of finding G. Folino (&) C. Mastroianni ICAR-CNR, Rendi, CS, Italy e-mail: folino@icar.cnr.it

Throughout the development of computing tools usage and computer science, from the calculating machines up and into the present day computers, the major technical problem to be solved was the reduction of the amount of components and their bulk. This led inevitably to the use of serial processing. Only quite recently we started mass producing and using multi-processing. Though the nature itself is infinitely parallelised, the technical problems of computer development led also to the development of such software tools and programming languages which mirrored the serial nature of computers, so that the serialisation of parallel natural processes is performed by humans, the programmers. However, modern day developments of scientific and every-day needs for computing power have led to the introduction of multiprocessors, GPUs, clusters, grids and clouds of computers, as to ascertain enough processing speed, power and memory for very complex algorithms. However, the human/computer interaction that supports these developments is still heavily based on “classical” computer programming languages, serial programming, and the multi-computer environment is accessed only through programmed interfaces, where all the burden of parallelising the, now well known, serial algorithms is again the job of (human) programmers. Virtue is a development primarily based on the idea - as there already is so much we know of mathematical, logical and other important basic algorithms used in many fields, and their computer implementation(s) - that we would be able to “raise” the level of computer “understanding” more towards the level of our own, human language communication. This means that, for example, mathematical operations in Virtue are performed not only on integers and reals (floats), but also on complex numbers, quaternions and octonions, and that they are freely intermixed. Or that all logical operations work also on multi-levelled and multi-dimensional logical values (and not just Boolean). These are mathematically well defined operations used very often in visualisations, as well as scientific modelling. Furthermore, another basic underlying idea of the development of Virtue is that the language developed may not destroy a possibly parallel structure of data, which then allows direct and user (human) independent automatic parallel execution. Within such an approach interactive sentences can be “decoded” (executed, performed, saved, used...) on a single processor, a multi-core, multiprocessor, GPU, cluster, grid or cloud system, or any combination of them. Designed on these basic principles, Virtue is a language which proposes a different approach, by keeping the inherent parallel structure of natural algorithms, and doing the parallel processing by itself, if it is algorithmically possible. Virtue is a syntactically very simple, yet semantically extremely complex language, offering no “reserved words”, synonyms, automation of memoisation, multiple word contexts, combined data types of anything Virtue supports (e.g. functions, symbol names, scalars, multidimensional sub-structured arrays etc.), stochastic processing, multivalued and multidimensional logic operations, multidimensional sub-structured file access structures, continuations etc. Therefore, due to this semantic richness and grammatical simpleness, in Virtue, for example, the text of the algorithm for Conway's “Game of Life” necessitates only 12 language tokens (7 words, 14 numbers in 3 vectors and 2 delimiters) in one sentence. 1 Further development of the idea allows for development of a more syntactically rich very high level human oriented machine interaction language, which, combined with additional artificial intelligence components, appropriate ergonomic human presentation/sensory interfaces and with the integration of user style association memory, we sincerely hope can help the future development of Computer Science and Usage Practice.