Summary The cloud computing paradigm has emerged as the backbone of modern price‐aware scalable computing systems. Many cloud service models are competing to become the leading doorway to access the computational power of cloud providers. Recently, a novel service model, called function‐as‐a‐service (FaaS), has been proposed, which enables users to exploit the cloud computational scalability, left out the configuration and management of huge computing infrastructures. This article discloses Fly, a domain‐specific language, which aims at reconciling cloud and high‐performance computing paradigms adopting a multicloud strategy by providing a powerful, effective, and pricing‐efficient tool for developing scalable workflow‐based scientific applications by exploiting different and at the same time FaaS cloud providers as computational backends in a transparent fashion. We present several improvements of the Fly language, as well as a new enhanced version of a source‐to‐source compiler, which currently supports Symmetric Multiprocessing, Amazon AWS, and Microsoft Azure backends and translation of functions in Java, JavaScript, and Python programming languages. Furthermore, we discuss a performance evaluation of Fly on a popular benchmark for distributed computing frameworks, along with a collection of case studies with an analysis of their performance results and costs.

Software that is meant to support collaboration is mostly developed “ad hoc”, placing some additional overhead to users, that are required to integrate the common work practices, realized with the traditional software applications, with the new collaborative features offered by the new application.

Several proposals have been presented to supplement the traditional measure of routing efficiency in P2P networks, i.e. the (average) number of hops for lookup operations, with measures of the latency incurred in the underlying network. So far, no solution has been presented to this “latency” problem without incurring in extra and heavy management costs. We propose Relaxed-2-Chord, a new design of the traditional Chord protocol, that is able to fit the routing tables with low latency nodes, doing a parasitic measurement of nodes' latency without adding any overhead. The solution that we present is a Distributed Hash Table system whose aim is to combine the routing efficiency and flexibility of the Chord protocol - i.e. a good degree/diameter tradeoff - and a provable optimal hop by hop latency. Our work is inspired by the recent Lookup-parasitic random sampling (LPRS) strategies which allow to improve the network stretch, that is, the ratio between the latency of two nodes on the overlay network and the unicast latency between those nodes. Relaxed-2-Chord reaches the same results as LPRS without introducing any overhead

Abstract The Chord protocol is the best known example of implementation of logarithmic complexity routing for structured peer‐to‐peer networks. Its routing algorithm, however, does not provide an optimal trade‐off between resources exploited (the size of the ‘finger table’) and performance (the average or worst‐case number of hops to reach destination). Cordasco et al. showed that a finger table based on Fibonacci distances provides lower number of hops with fewer table entries. In this paper we generalize this result, showing how to construct an improved finger table when the objective is to reduce the number of hops, possibly at the expense of an increased size of the finger table. Our results can also be exploited to guarantee low routing time in case a fraction of nodes fails. Copyright © 2007 John Wiley & Sons, Ltd.

Abstract We propose a family of novel Chord‐based P2P schemes retaining all positive aspects that made Chord a popular topology for routing in P2P networks. The schemes, based on the Fibonacci number system, allow to simultaneously improve on the maximum/average number of hops for lookups and the routing table size per node. © 2008 Wiley Periodicals, Inc. NETWORKS, 2008

We define a proximity overlay network (PON) which allow to realize DHT systems whose aim is to combine routing efficiency - i.e. an optimal degree/diameter tradeoff - and proximity awareness. The proposed systems is parameterized with a positive integer s which measures the amount of flexibility offered by the network. Varying the value of s the system goes from a quite rigid network (s=2) which offer an optimal degree/diameter tradeoff. Increasing s to relatively low values allows to increase the flexibility of the network and consequently improves the stretch, that is, the ratio between the latency of two nodes on the overlay network and the unicast latency between those nodes. We are able to reconcile the conflict between the load balancing and proximity relationship by proving the efficiency of the main performance metrics. In particular we analytically prove that our system can result in lookup latencies proportional to the maximum latency of the underlying physical network, provided that the physical network has a power law latency expansion.

The chord protocol is the best known example of implementation of logarithmic complexity routing for structured peer-to-peer networks. Its routing algorithm, however, does not provide an optimal trade-off between resources exploited (the size of the "finger table") and performance (the average or worst-case number of hops to reach destination). Cordasco et al. showed that a finger table based on Fibonacci distances provides lower number of hops with fewer table entries. In this paper, we generalize this result, showing how to construct an improved finger table when the objective is to reduce the number of hops, possibly at the expense of an increased size of the finger table. Our results can also be exploited to guarantee low routing time in case a fraction of nodes is assumed to fail.

Electronic Commerce (EC) is currently one of the fastest growing and most practically relevant application areas of distributed systems technologies. It is based on the economic aspects of commercial trading patterns combined with distributed computing systems technology. It is a market environment that is characterized by low transaction costs, a large number of market participants, and easy online access to services and goods offered. It also implies a set of rules and policies for the successful organization of business transactions. EC involves more than simple online transactions, it encompasses diverse activities as conducting market research, identifying opportunities and partners, cultivating relationships with customers and suppliers, document exchange and customer modeling. Our paper deals with the latter aspect of EC. We introduce here a model for developing a symmetric adaptive system for EC on the World Wide Web. Our main contribution is that the model is, by all means, symmetric: we model both customers and goods and make both their profiles change as a consequence of a customer buying a certain product. The symmetry in our model greatly simplifies the approach and the queries, giving some insights on the formalization of the allowed queries that were, in way, unexpected. Furthermore, the model itself can provide an easy-to-evaluate measure for the confidence in adapting its response to any given customer and is able to provide useful feedback to the manager, then allowing, so to speak, "manual adjustment" that can help the behaviour of the system in the future.

We consider the problem of selecting the Kth smallest element of a set distributed among the sites of a communication network when the size of messages is bounded; that is, each message is a packet which contains at most c bits, where c/spl ges/1 is a constant. A general selection algorithm using packets is presented and its packet complexity is analyzed. Its complexity is shown to be a significant improvement for a large range of packet sizes over the existing bounds. The proposed technique is then instanciated for specific classes of network topologies; the resulting bounds either match or improve the ones of existing solutions for a large range of values of the packet size. Furthermore, it is bit optimal in star networks.

One-and two-dimensional pattern matching oriented systolic array processors are presented that support, respectively, the detection of all repetitions in a string x and the statistics of all substrings of x with and without overlap. The time is linear in the length of x in both applications, whereas the number of processors is linear and quadratic, respectively.