Multi-server systems have received increasing attention with important implementations such as Google MapReduce, Hadoop, and Spark. Common to these systems are a fork operation, where jobs are first divided into tasks that are processed in parallel, and a later join operation, where completed tasks wait until the results of all tasks of a job can be combined and the job leaves the system. The synchronization constraint of the join operation makes the analysis of fork-join systems challenging and few explicit results are known. In this work, we model fork-join systems using a max-plus server model that enables us to derive statistical bounds on waiting and sojourn times for general arrival and service time processes. We contribute end-to-end delay bounds for multi-stage fork-join networks that grow in O(h ln k) for h fork-join stages, each with k parallel servers. We perform a detailed comparison of different multiserver configurations and highlight their pros and cons. We also include an analysis of single-queue fork-join systems that are nonidling and achieve a fundamental performance gain, and compare these results to both simulation and a live Spark system.
Mobile Wireless Delay-Tolerant Networks (DTNs) are wireless networks that suffer from intermittent connectivity, but enjoy the benefit of mobile nodes that can store and forward packets or messages, and can act as relays, bringing packets and messages closer to their destination through a selective forwarding policy. Many DTN protocols compensate for the unpredictability of the network by distributing multiple message copies in the hopes that at least one will eventually be delivered. As the number of message carriers becomes large these schemes experience diminishing marginal benefits from the addition of more message carriers. We describe and analyze the Simple Counting Protocol, an extremely simple and robust method for limiting the fraction of nodes that carry a copy of a message. We examine the performance of this protocol in conjunction with several abstract mobility models and show that the protocol performs reasonably well in diverse circumstances. The Simple Counting Protocol does not assume much about node mobility, and therefore should be useful for applications where little is known about node encounter patterns. The simplicity of its implementation will hopefully make it a useful substitute for epidemic routing as a naive lower bound in protocol performance comparisons.We also show how the same simple techniques and principles can be applied in conjunction with more complex heuristic DTN protocols to reduce network resource usage, a scheme we call Intermediate Immunity.
In order to better understand human and animal mobility and its potential effects on Mobile Ad-Hoc networks and Delay-Tolerant Networks, many researchers have conducted experiments which collect encounter data. Most analyses of these data have focused on isolated statistical properties such as the distribution of node inter-encounter times and the degree distribution of the connectivity graph.On the other hand, new developments in computational topology, in particular persistent homology, have made it possible to compute topological invariants from noisy data. These homological methods provide a natural way to draw conclusions about global structure based on collections of local information.We use persistent homology techniques to show that in some cases encounter traces can be used to deduce information about the topology of the physical space the experiment was conducted in, and detect certain changes in the space. We also show that one can distinguish between simulated encounter traces generated on a bounded rectangular grid from traces generated on a grid with the opposite edges wrapped (a toroidal grid). Finally, we have found that nontrivial topological features also appear in real experimental encounter traces, and we speculate on types of node behavior that could produce these results. This demonstrates the ability of persistent homology to detect topological features in encounter data that could be difficult to describe using traditional statistical and geometric methods. Figure 1: The topology of the space affects the type of encounter patterns that are possible. If the space is like a line, node A cannot encounter node C without one of them encountering node B. If the space is like a loop, node A and C can encounter each other without encountering node B.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.