Results 1 
8 of
8
Modular Competitiveness for Distributed Algorithms
 In Proc. 28th ACM Symp. on Theory of Computing (STOC
, 2000
"... We define a novel measure of competitive performance for distributed algorithms based on throughput, the number of tasks that an algorithm can carry out in a fixed amount of work. This new measure complements the latency measure of Ajtai et al. [3], which measures how quickly an algorithm can finish ..."
Abstract

Cited by 14 (2 self)
 Add to MetaCart
(Show Context)
We define a novel measure of competitive performance for distributed algorithms based on throughput, the number of tasks that an algorithm can carry out in a fixed amount of work. This new measure complements the latency measure of Ajtai et al. [3], which measures how quickly an algorithm can finish tasks that start at specified times. An important property of the throughput measure is that it is modular: we define a notion of relative competitiveness with the property that a krelatively competitive implementation of an object T using a subroutine U , combined with an lcompetitive implementation of U , gives a klcompetitive algorithm for ...
Approximation Algorithms for Data Management in Networks
, 2001
"... This paper deals with static data management in computer systems connected by networks. A basic functionality in these systems is the interactive ..."
Abstract

Cited by 14 (0 self)
 Add to MetaCart
This paper deals with static data management in computer systems connected by networks. A basic functionality in these systems is the interactive
A Data Tracking Scheme for General Networks
, 2001
"... Consider an arbitrary distributed network in which large numbers of objects are continuously being created, replicated, and destroyed. A basic problem arising in such an environment is that of organizing a distributed directory service for locating object copies. In this paper, we present a new data ..."
Abstract

Cited by 13 (0 self)
 Add to MetaCart
Consider an arbitrary distributed network in which large numbers of objects are continuously being created, replicated, and destroyed. A basic problem arising in such an environment is that of organizing a distributed directory service for locating object copies. In this paper, we present a new data tracking scheme for locating nearby copies of objects in arbitrary distributed environments. Our tracking scheme supports ecient accesses to data objects while keeping the local memory overhead low. In particular, our tracking scheme achieves an expected polylog(n) approximation in the cost of any access operation, for an arbitrary network. The memory overhead incurred by our scheme is O(polylog(n)) times the maximum number of objects stored at any node, with high probability. We also show that our tracking scheme adapts well to dynamic changes in the network.
New Results for Online Page Replication
, 2003
"... We study the online page replication problem. We present a new randomized online algorithm for rings which is 2.37297competitive, improving the best previous result of 3.16396. We also show that no randomized online algorithm is better than 1.75037competitive on the ring; previously, only a 1.5819 ..."
Abstract

Cited by 6 (0 self)
 Add to MetaCart
We study the online page replication problem. We present a new randomized online algorithm for rings which is 2.37297competitive, improving the best previous result of 3.16396. We also show that no randomized online algorithm is better than 1.75037competitive on the ring; previously, only a 1.58198 bound for a single edge was known. We extend the problem in several new directions: continuous metrics, variable size requests, and replication before service. This yields simplified proofs of several known results.
Competitive Algorithms for Asynchronous Update Propagation in Mobile Computing and for Search Engine Freshness
"... Mobile computing has become a hot area of research because of proliferation of laptop computers. Some of the challenges in using wireless networks for mobile computing are low bandwidth and high communication cost. To address these challenges, researchers have built systems in which the file system ..."
Abstract
 Add to MetaCart
Mobile computing has become a hot area of research because of proliferation of laptop computers. Some of the challenges in using wireless networks for mobile computing are low bandwidth and high communication cost. To address these challenges, researchers have built systems in which the file system contents on the backbone file server are optimistically replicated on the laptop. The problem of how often to send the updates from the laptop to the backbone server for optimistically replicated filesystems remains unsolved. Search engine freshness has become an important problems with the emergence of search engines which cache a copy of the document. Such a cache can provide valuable fault tolerance when the original web site is unreachable. Because of the growing use of laptops and search engines, maintaining freshness of data is becoming an important problem. There is very sparse literature on solving the problem of freshness. In this paper, we introduce a cost model which takes...
Distributed and OnLine Routing on Tori
, 1997
"... In this paper, we study the problem of assigning paths to packets on N N tori in an online and distributed fashion. By online we mean that the routing decisions must be made without any knowledge of future requests. Being distributed is an equally important feature of our design, for such algorith ..."
Abstract
 Add to MetaCart
(Show Context)
In this paper, we study the problem of assigning paths to packets on N N tori in an online and distributed fashion. By online we mean that the routing decisions must be made without any knowledge of future requests. Being distributed is an equally important feature of our design, for such algorithms need not know the global configuration of the network in the process of routing packets. We use the technique of competitive analysis to measure the performance of our design. In addition to showing an log N) lower bound on the competitive ratio, we present both deterministic and randomized algorithms which are O(log N) competitive with respect to the maximum load (i.e., congestion) on communication links.
Competitive Analysis of Oblivious Routing on Hypercubes
, 2000
"... We study the performance of oblivious routing algorithms using competitive analysis on a d dimensional, N = 2 d node hypercube. Packets are injected into the hypercube arbitrarily and continuously, without any (e.g., probabilistic) assumption on the arrival of packets. First we consider oblivious ..."
Abstract
 Add to MetaCart
We study the performance of oblivious routing algorithms using competitive analysis on a d dimensional, N = 2 d node hypercube. Packets are injected into the hypercube arbitrarily and continuously, without any (e.g., probabilistic) assumption on the arrival of packets. First we consider oblivious routing algorithms that follow minimal (shortest) paths. We show that the well known deterministic oblivious routing algorithm, i.e., the greedy routing algorithm, has competitive ratio N 1=2 ). However, a randomized variant of the greedy routing algorithm is shown to be only in a factor of O(log 3 N) to our supplied problem lower bound, which is of order N log 2 (5=4) = log 5 N ), for any minimal randomized oblivious routing algorithms. The lower bound we provide is strong in that the adversaries also follow minimal paths. We do not relay on the well known fact that any permutation can be routed with constant congestion, which in general needs nonminimal paths. Next we conside...
Fighting Against Two Adversaries: Page Migration in Dynamic Networks (extended abstract)
"... Page migration is one of the fundamental subproblems in the framework of data management in networks. It occurs in a distributed network of processors sharing one indivisible memory page of size D, which is stored in one of the processors. During runtime, processors access unit size data items from ..."
Abstract
 Add to MetaCart
Page migration is one of the fundamental subproblems in the framework of data management in networks. It occurs in a distributed network of processors sharing one indivisible memory page of size D, which is stored in one of the processors. During runtime, processors access unit size data items from the page, and the system is allowed to move the page from one processor to another in order to minimize the total communication cost. This problem was considered in the online setting numerous times by many researchers, and some online algorithms were proven to achieve a cost within a constant factor of the optimal oine solution. However, all results were achieved under the assumption that the communication costs between processors were xed during the execution of the whole process. In this paper we consider a model in which the communication costs can change in each time step, but the pace of the changes is restricted. This is typical in mobile networks, and also models the dynamics of networks that are not exclusively dedicated to the page migration. If both changes of the network and the request sequence are given by some adversarial entity, we prove a tight bound on the competitive ratio of the problem. However, the size of this ratio motivates us to assume that the changes of communication costs are modeled by some stochastic process, and an adversary dictates only which processor issues The full version of this paper is available under