Results 1  10
of
494
The Power of Two Choices in Randomized Load Balancing
 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
, 1996
"... Suppose that n balls are placed into n bins, each ball being placed into a bin chosen independently and uniformly at random. Then, with high probability, the maximum load in any bin is approximately log n log log n . Suppose instead that each ball is placed sequentially into the least full of d ..."
Abstract

Cited by 292 (23 self)
 Add to MetaCart
Suppose that n balls are placed into n bins, each ball being placed into a bin chosen independently and uniformly at random. Then, with high probability, the maximum load in any bin is approximately log n log log n . Suppose instead that each ball is placed sequentially into the least full of d bins chosen independently and uniformly at random. It has recently been shown that the maximum load is then only log log n log d +O(1) with high probability. Thus giving each ball two choices instead of just one leads to an exponential improvement in the maximum load. This result demonstrates the power of two choices, and it has several applications to load balancing in distributed systems. In this thesis, we expand upon this result by examining related models and by developing techniques for stu...
Algorithms for Parallel Memory I: TwoLevel Memories
, 1992
"... We provide the first optimal algorithms in terms of the number of input/outputs (I/Os) required between internal memory and multiple secondary storage devices for the problems of sorting, FFT, matrix transposition, standard matrix multiplication, and related problems. Our twolevel memory model is n ..."
Abstract

Cited by 235 (27 self)
 Add to MetaCart
(Show Context)
We provide the first optimal algorithms in terms of the number of input/outputs (I/Os) required between internal memory and multiple secondary storage devices for the problems of sorting, FFT, matrix transposition, standard matrix multiplication, and related problems. Our twolevel memory model is new and gives a realistic treatment of parallel block transfer, in which during a single I/O each of the P secondary storage devices can simultaneously transfer a contiguous block of B records. The model pertains to a largescale uniprocessor system or parallel multiprocessor system with P disks. In addition, the sorting, FFT, permutation network, and standard matrix multiplication algorithms are typically optimal in terms of the amount of internal processing time. The difficulty in developing optimal algorithms is to cope with the partitioning of memory into P separate physical devices. Our algorithms' performance can be significantly better than those obtained by the wellknown but nonopti...
A VehicletoVehicle Communication Protocol for Cooperative Collision Warning
, 2004
"... This paper proposes a vehicletovehicle communication protocol for cooperative collision warning. Emerging wireless technologies for vehicletovehicle (V2V) and vehicletoroadside (V2R) communications such as DSRC [1] are promising to dramatically reduce the number of fatal roadway accidents by pr ..."
Abstract

Cited by 149 (0 self)
 Add to MetaCart
(Show Context)
This paper proposes a vehicletovehicle communication protocol for cooperative collision warning. Emerging wireless technologies for vehicletovehicle (V2V) and vehicletoroadside (V2R) communications such as DSRC [1] are promising to dramatically reduce the number of fatal roadway accidents by providing early warnings. One major technical challenge addressed in this paper is to achieve lowlatency in delivering emergency warnings in various road situations. Based on a careful analysis of application requirements, we design an effective protocol, comprising congestion control policies, service differentiation mechanisms and methods for emergency warning dissemination. Simulation results demonstrate that the proposed protocol achieves low latency in delivering emergency warnings and efficient bandwidth usage in stressful road scenarios. 1.
Efficient replica maintenance for distributed storage systems
 In Proc. of NSDI
, 2006
"... This paper considers replication strategies for storage systems that aggregate the disks of many nodes spread over the Internet. Maintaining replication in such systems can be prohibitively expensive, since every transient network or host failure could potentially lead to copying a server’s worth of ..."
Abstract

Cited by 122 (17 self)
 Add to MetaCart
(Show Context)
This paper considers replication strategies for storage systems that aggregate the disks of many nodes spread over the Internet. Maintaining replication in such systems can be prohibitively expensive, since every transient network or host failure could potentially lead to copying a server’s worth of data over the Internet to maintain replication levels. The following insights in designing an efficient replication algorithm emerge from the paper’s analysis. First, durability can be provided separately from availability; the former is less expensive to ensure and a more useful goal for many widearea applications. Second, the focus of a durability algorithm must be to create new copies of data objects faster than permanent disk failures destroy the objects; careful choice of policies for what nodes should hold what data can decrease repair time. Third, increasing the number of replicas of each data object does not help a system tolerate a higher disk failure probability, but does help tolerate bursts of failures. Finally, ensuring that the system makes use of replicas that recover after temporary failure is critical to efficiency. Based on these insights, the paper proposes the Carbonite replication algorithm for keeping data durable at a low cost. A simulation of Carbonite storing 1 TB of data over a 365 day trace of PlanetLab activity shows that Carbonite is able to keep all data durable and uses 44 % more network traffic than a hypothetical system that only responds to permanent failures. In comparison, Total Recall and DHash require almost a factor of two more network traffic than this hypothetical system. 1
Capprobe: A simple and accurate capacity estimation technique
 In Proceeding ACM SIGCOMM
, 2004
"... We present a new capacity estimation technique, called CapProbe. CapProbe combines delay as well as dispersion measurements of packet pairs to filter out samples distorted by crosstraffic. CapProbe algorithms include convergence tests and convergence speedup techniques by varying probing parame ..."
Abstract

Cited by 115 (25 self)
 Add to MetaCart
(Show Context)
We present a new capacity estimation technique, called CapProbe. CapProbe combines delay as well as dispersion measurements of packet pairs to filter out samples distorted by crosstraffic. CapProbe algorithms include convergence tests and convergence speedup techniques by varying probing parameters. Our study of CapProbe includes a probability analysis to determine the time it takes CapProbe to converge on the average. Through simulations and measurements, we found CapProbe to be quick and accurate across a wide range of traffic scenarios. We also compared CapProbe with two previous wellknown techniques, pathchar and pathrate. We found CapProbe to be much more accurate than pathchar and similar in accuracy to pathrate, while providing faster estimation than both. Another advantage of CapProbe is its lower computation cost, since no statistical post processing of probing data is required.
Classifying scheduling policies with respect to unfairness in an M/GI/1
 Proc. of SIGMETRICS’03
, 2003
"... It is common to classify scheduling policies based on their mean response times. Another important, but sometimes opposing, performance metric is a scheduling policy’s fairness. For example, a policy that biases towards short jobs so as to minimize mean response time, may end up being unfair to long ..."
Abstract

Cited by 96 (17 self)
 Add to MetaCart
It is common to classify scheduling policies based on their mean response times. Another important, but sometimes opposing, performance metric is a scheduling policy’s fairness. For example, a policy that biases towards short jobs so as to minimize mean response time, may end up being unfair to long jobs. In this paper we define three types of unfairness and demonstrate large classes of scheduling policies that fall into each type. We end with a discussion on which jobs are the ones being treated unfairly. 1
Selfish Traffic Allocation for Server Farms
, 2003
"... We study the price of selfish routing in noncooperative networks like the Internet. In particular, we investigate the price... ..."
Abstract

Cited by 76 (5 self)
 Add to MetaCart
We study the price of selfish routing in noncooperative networks like the Internet. In particular, we investigate the price...
An Analytic Behavior Model for Disk Drives With Readahead Caches and Request Reordering
, 1998
"... Modern disk drives readahead data and reorder incoming requests in a workloaddependent fashion. This improves their performance, but makes simple analytical models of them inadequate for performance prediction, capacity planning, workload balancing, and so on. To address this problem we have devel ..."
Abstract

Cited by 74 (8 self)
 Add to MetaCart
Modern disk drives readahead data and reorder incoming requests in a workloaddependent fashion. This improves their performance, but makes simple analytical models of them inadequate for performance prediction, capacity planning, workload balancing, and so on. To address this problem we have developed a new analytic model for disk drives that do readahead and request reordering. We did so by developing performance models of the disk drive components (queues, caches, and the disk mechanism) and a workload transformation technique for composing them. Our model includes the effects of workloadspecific parameters such as request size and spatial locality. The result is capable of predicting the behavior of a variety of realworld devices to within 17% across a variety of workloads and disk drives.
Tuple Routing Strategies for Distributed Eddies
, 2003
"... Many applications that consist of streams of data are inherently distributed. Since input stream rates and other system parameters such as the amount of available computing resources can fluctuate significantly, a stream query plan must be able to adapt to these changes. Routing tuples between ..."
Abstract

Cited by 71 (2 self)
 Add to MetaCart
Many applications that consist of streams of data are inherently distributed. Since input stream rates and other system parameters such as the amount of available computing resources can fluctuate significantly, a stream query plan must be able to adapt to these changes. Routing tuples between operators of a distributed stream query plan is used in several data stream management systems as an adaptive query optimization technique. The routing policy used can have a significant impact on system performance. In this paper, we use a queuing network to model a distributed stream query plan and define performance metrics for response time and system throughput. We also propose and evaluate several practical routing policies for a distributed stream management system. The performance results of these policies are compared using a discrete event simulator.