In this paper we study the space requirement of algorithms that make only one (or a small number of) pass(es) over the input data. We study such algorithms under a model of data streams that we introduce here. We give a number of upper and lower bounds for problems stemming from queryprocessing, invoking in the process tools from the area of communication complexity.

AbstractÐWe consider the problem of load balancing in dynamic distributed systems in cases where new incoming tasks can make use of old information. For example, consider a multiprocessor system where incoming tasks with exponentially distributed service requirements arrive as a Poisson process, the tasks must choose a processor for service, and a task knows when making this choice the processor queue lengths from T seconds ago. What is a good strategy for choosing a processor in order for tasks to minimize their expected time in the system? Such models can also be used to describe settings where there is a transfer delay between the time a task enters a system and the time it reaches a processor for service. Our models are based on considering the behavior of limiting systems where the number of processors goes to infinity. The limiting systems can be shown to accurately describe the behavior of sufficiently large systems and simulations demonstrate that they are reasonably accurate even for systems with a small number of processors. Our studies of specific models demonstrate the importance of using randomness to break symmetry in these systems and yield important rules of thumb for system design. The most significant result is that only small amounts of queue length information can be extremely useful in these settings; for example, having incoming tasks choose the least loaded of two randomly chosen processors is extremely effective over a large range of possible system parameters. In contrast, using global information can actually degrade performance unless used carefully; for example, unlike most settings where the load information is current, having tasks go to the apparently least loaded server can significantly hurt performance. Index TermsÐLoad balancing, stale information, old information, queuing theory, large deviations. æ 1

In this paper we examine the problem of balancing load in a large-scale distributed system when information about server loads may be stale. It is well known that sending each request to the machine with the apparent lowest load can behave badly in such systems, yet this technique is common in practice. Other systems use round-robin or random selection algorithms that entirely ignore load information or that only use a small subset of the load information. Rather than risk extremely bad performance on one hand or ignore the chance to use load information to improve performance on the other, we develop strategies that interpret load information based on its age. Through simulation, we examine several simple algorithms that use such load interpretation strategies under a range of workloads. Our experiments suggest that by properly interpreting load information, systems can (1) match the performance of the most aggressive algorithms when load information is fresh relative to the...

Let X = f1; 2; : : : ; ng be a ground set of n elements, and let S be a family of subsets of X , jSj = m, with a positive cost c S associated with each S 2 S.

We consider the problem of scheduling jobs that are given as groups of non-intersecting segments on the real line. Each job J j is associated with an interval, I j , which consists of up to t segments, for some t 1, a positive weight, w j , and two jobs are in conflict if any of their segments intersect. Such jobs show up in a wide range of applications, including the transmission of continuous-media data, allocation of linear resources (e.g. bandwidth in linear processor arrays), and in computational biology/geometry. The objective is to schedule a subset of non-conflicting jobs of maximum total weight.

We formulate the problem of Video on Demand (VOD) from a resource allocation perspective. We introduce the decision element into a movie vending environment, which complements the current approaches. In contrast with the traditional resource allocation problems (e.g., machine scheduling), the problem possesses the distinctive batching property, which stands for the feasibility of several requests being served by one resource. First, we investigate the problem in an on-line fashion, namely, having to accept or reject a request for a movie without any knowledge about future requests. We show upper and lower bounds on the competitive ratio of on-line movie scheduling algorithms for a variety of scenarios depending on the notification time. In particular, when the length of the movie and the notification time on requests are linearly related, we present a simple competitive algorithm. Second, we propose a new approach -- Shared VOD -- which gives a more balanced functionality /p...

this paper we have described competitive on-line algorithms for on-line network routing problems. We have concentrated on routing in electrical and optical networks, presented algorithms for load minimization and throughput maximization problems, and mentioned some of the most popular open problems in the area.

. Most of the machines that are connected to the Internet are idle a significant fraction of the time. This paper presents the Java Market, a system that allows organizations and Internet users to make use of this wasted computational power. Using the Java programming language and the Web technology, the Java Market is the first metacomputing system that can seamlessly take advantage of machines of any architecture, anywhere on the Internet. Every user on the Internet can contribute their machine's computational resources just by pointing a Java-capable browser to the Java Market web page. Similarly, users can launch jobs to the system by posting them on the Web and registering them with the Java Market. In contrast, other systems that allow sharing of computational resources between machines over the network require homogeneous system architecture. They often involve extensive installation or even kernel-level modifications to the operating system. 1. Introduction The wasted CPU cy...

This paper studies the problem of balancing the demand for content in a peer-to-peer network across heterogeneous peer nodes that hold replicas of the content. Previous decentralized load balancing techniques in distributed systems base their decisions on periodic updates containing information about load or available capacity observed at the serving entities. We show that these techniques do not work well in the peer-to-peer context; either they do not address peer node heterogeneity, or they suffer from significant load oscillations which result in unutilized capacity. We propose a new decentralized algorithm, Max-Cap, based on the maximum inherent capacities of the replica nodes. We show that unlike previous algorithms, it is not tied to the timeliness or frequency of updates, and consequently requires significantly less update overhead. Yet, Max-Cap can handle the heterogeneity of a peer-to-peer environment without suffering from load oscillations.

Multicast routing and admission control can be defined as follows. Various network users issue online requests for a connection to certain multicast sources. The network either connects this request or rejects it (admission control decision). If request is connected, a path with sufficient bandwidth is established starting at the requesting user and ending either at the source or at one of the users previously connected to the same source (route selection decision). The goal of these admission control and route selection decisions is to maximize total number of users connected (i.e. total thruput) subject to the network capacity constraints. This problem can be reduced to the online maximal dense tree problem: Upon a service request from a node in a weighted graph with a distinguished source, the node is either rejected or connected to the tree of previously connected nodes. The goal is to maximize the total number of requests while keeping the density (ratio of accepted re...