Minimizing Multicast Communication Latency in WAN-Based Environment Silawarawet, K.; Nupairoj, N.
Seventh ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing (SNPD'06),
2006
Conference Proceeding
The critical problem for multicasting is to find optimal multicast tree. Most popular multicast algorithms over WAN are usually based on either Dijkstra's shortest path tree or Steiner tree, which ...assume that sender can send a message to the next destination node right away. This can lead to a non-optimal multicast tree as it ignores gap parameter in LogP model. From our preliminary study, we found that gap is very significant especially in WAN. In this paper, we take gap into consideration and propose a minimization technique, which apply gap parameters to Dijkstra's shortest path algorithm to construct minimized shortest path trees. We compare the multicast latency of normal Dijkstra's shortest path multicast tree and our minimized multicast tree with simulation. The results show that multicast tree based on our proposed algorithm can clearly out-perform the multicast tree based on Dijkstra's algorithm. Our results further indicate that our proposed algorithm can perform very well when message size is large with lots of nodes
Grid technology has been proposed as a computing framework for gaining access to very large distributed computing resources. In order to fully utilize these resources, we need load sharing policy ...that can address the unique characteristics in grid infrastructure. In this paper, we extend our implicit load sharing policy by introducing the movable master node concept to hide an effect of the high latency in grid computing environment. The simulation results show that our proposed policy can achieve better parallel runtime in grid computing environment by moving the job pool to the proper master node
A key issue in designing software multicast algorithms is to consider the trade-off between performance and portability. Portable software multicast algorithms based on generic communication models ...cannot capture some architecture-specific features. Without considering the underlying network architecture, these multicast algorithms may not achieve the truly optimal performance when implemented in real networks. The objective of this research is to investigate architecture-dependent tuning on performance of multicast algorithms developed based on architecture-independent models. Specifically, we intend to optimize the multicast algorithm based on the parameterized communication model. We propose two multicast algorithms, OPT-mesh and OPT-min which are the optimized versions of the parameterized multicast algorithm for wormhole-switched mesh networks and BMIN networks, respectively. Using our flit-level simulator the performance of both algorithms are compared with the architecture-independent version of the parameterized multicast algorithm and two other well-known network-dependent algorithms based on the binomial tree.
Message Passing Interface (MPI) is an attempt to standardize the communication library for distributed-memory computing systems. Since the release of the recent MPI specification, several MPI ...implementations have been made publicly available. Different implementations employ different approaches, and thus, the performance of each implementation may vary. Since the performance of communication is extremely crucial to message-passing based applications, selecting an appropriate MPI implementation becomes critical. Our study is intended to provide a guideline on how to perform such a task on workstation clusters which are known to be an economical and effective platform in high performance computing. We investigate several MPI aspects including its functionalities and performance. Our results also point out the strength and weakness of each implementation on our experimental system.< >