Results 11  20
of
82
Patterns Of Buffer Overflow In A Class Of Queues With Long Memory In The Input Stream
 Ann. Appl. Probab
, 1997
"... . We study the time it takes until a a fluid queue with a finite, but large, holding capacity reaches the overflow point. The queue is fed by an on/off process, with a heavy tailed on distribution, which is known to have long memory. It turns out that the expected time until overflow, as a function ..."
Abstract

Cited by 35 (9 self)
 Add to MetaCart
. We study the time it takes until a a fluid queue with a finite, but large, holding capacity reaches the overflow point. The queue is fed by an on/off process, with a heavy tailed on distribution, which is known to have long memory. It turns out that the expected time until overflow, as a function of capacity L, increases only polynomially fast, and so overflows happen much more often than in the "classical" light tailed case, where the expected overflow time increases as an exponential function of L. Moreover, we show that in the heavy tailed case, overflows are basically caused by single huge jobs. An implication is that the usual GI=G=1 queue with finite but large holding capacity and heavy tailed service times, will overflow about equally often no matter how much we increase the service rate. We also study the time until overflow for queues fed by a superposition of k iid. on/off processes with a heavy tailed on distribution, and show the benefit of pooling the system resources a...
Characterizing Network Traffic in a Clusterbased, Multitier Data Center
 In Proceedings of the 27th International Conference on Distributed Computing Systems (ICDCS
"... With the increasing use of various Webbased services, design of high performance, scalable and dependable data centers has become a critical issue. Recent studies show that a clustered, multitier architecture is a costeffective approach to design such servers. Since these servers are highly distr ..."
Abstract

Cited by 26 (4 self)
 Add to MetaCart
(Show Context)
With the increasing use of various Webbased services, design of high performance, scalable and dependable data centers has become a critical issue. Recent studies show that a clustered, multitier architecture is a costeffective approach to design such servers. Since these servers are highly distributed and complex, understanding the workloads driving them is crucial for the success of the ongoing research to improve them. In view of this, there has been a significant amount of work to characterize the workloads of Webbased services. However, all of the previous studies focus on a high level view of these servers, and analyze requestbased or sessionbased characteristics of the workloads. In this paper, we focus on the characteristics of the network behavior within a clustered, multitiered data center. Using a real implementation of a clustered threetier data center, we analyze the arrival rate and interarrival time distribution of the requests to individual server nodes, the network traffic between tiers, and the average size of messages exchanged between tiers. The main results of this study are; (1) in most cases, the request interarrival rates follow lognormal distribution, and selfsimilarity exists when the data center is heavily loaded, (2) message sizes can be modeled by the lognormal distribution, and (3) service times fit reasonably well with the Pareto distribution and show heavy tailed behavior at heavy loads. 1.
THE BARISTA: A MODEL FOR BID ARRIVALS IN ONLINE AUCTIONS
, 712
"... The arrival process of bidders and bids in online auctions is important for studying and modeling supply and demand in the online marketplace. A popular assumption in the online auction literature is that a Poisson bidder arrival process is a reasonable approximation. This approximation underlies th ..."
Abstract

Cited by 20 (6 self)
 Add to MetaCart
(Show Context)
The arrival process of bidders and bids in online auctions is important for studying and modeling supply and demand in the online marketplace. A popular assumption in the online auction literature is that a Poisson bidder arrival process is a reasonable approximation. This approximation underlies theoretical derivations, statistical models and simulations used in field studies. However, when it comes to the bid arrivals, empirical research has shown that the process is far from Poisson, with early bidding and lastmoment bids taking place. An additional feature that has been reported by various authors is an apparent selfsimilarity in the bid arrival process. Despite the wide evidence for the changing bidding intensities and the selfsimilarity, there has been no rigorous attempt at developing a model that adequately approximates bid arrivals and accounts for these features. The goal of this paper is to introduce a family of distributions that wellapproximate the bid time distribution in hardclose auctions.
A Performance Study of Internet Web Servers
"... The phenomenal growth in popularity of the WorldWide Web (WWW or "the Web") has made WWW traffic the fastest growing component of network packet and byte traffic on the NSFNET backbone. This growth has triggered much recent research activity aimed at improving Web performance and scalabil ..."
Abstract

Cited by 18 (1 self)
 Add to MetaCart
The phenomenal growth in popularity of the WorldWide Web (WWW or "the Web") has made WWW traffic the fastest growing component of network packet and byte traffic on the NSFNET backbone. This growth has triggered much recent research activity aimed at improving Web performance and scalability: reducing the volume of network traffic produced by Web clients and servers by using file/document caching, and reducing the latency for WWW users by using improved protocols for Web interaction. Fundamental to the goal of improving WWW performance is an understanding of WWW workloads. This thesis presents a workload characterization study for WorldWide Web servers. Six different data sets are used in this study: three from academic (i.e., university) environments, two from scientific research organizations, and one from a commercial Internet provider. These data sets represent three different orders of magnitude in server activity, and two different orders of magnitude in time duration, ranging...
Why NonLinearities Can Ruin The Heavy Tailed Modeler's Day
"... . A heavy tailed time series that can be expressed as an infinite order moving average has the property that the sample autocorrelation function (acf) at lag h, converges in probability to a constant ae(h) despite the fact that the mathematical correlation typically does not exist. A simple bilinear ..."
Abstract

Cited by 17 (8 self)
 Add to MetaCart
(Show Context)
. A heavy tailed time series that can be expressed as an infinite order moving average has the property that the sample autocorrelation function (acf) at lag h, converges in probability to a constant ae(h) despite the fact that the mathematical correlation typically does not exist. A simple bilinear model considered by Davis and Resnick (1996) has the property that the sample autocorrelation function at lag h converges in distribution to a nondegenerate random variable. Examination of various data sets exhibiting heavy tailed behavior reveals that the sample correlation function typically does not behave like a constant. Usually, the sample acf of the first half of the data set looks considerably different than the sample acf of the second half. A possible explanation for this acf behavior is the presence of nonlinear components in the underlying model and this seems to imply that infinite order moving average models and in particular ARMA models do not adequately capture dependency s...
SelfSimilar Communication Models And Very Heavy Tails
, 1998
"... Several studies of file sizes either being downloaded or stored in the world wide web have commented that tails can be so heavy that not only are variances infinite, but so are means. Motivated by this fact, we study the infinite node Poisson model under the assumption that transmission times are h ..."
Abstract

Cited by 17 (5 self)
 Add to MetaCart
Several studies of file sizes either being downloaded or stored in the world wide web have commented that tails can be so heavy that not only are variances infinite, but so are means. Motivated by this fact, we study the infinite node Poisson model under the assumption that transmission times are heavy tailed with infinite mean. The model is unstable but we are able to provide growth rates. Selfsimilar but nonstationary Gaussian process approximations are provided for the number of active sources, cumulative input, buffer content and time to buffer overflow.
Empirical Testing Of The Infinite Source Poisson Data Traffic Model
, 2000
"... The infinite source Poisson model is a fluid queue approximation of network data transmission that assumes that sources begin constant rate transmissions of data at Poisson time points for random lengths of time. This model has been a popular one as analysts attempt to provide explanations for obser ..."
Abstract

Cited by 16 (4 self)
 Add to MetaCart
(Show Context)
The infinite source Poisson model is a fluid queue approximation of network data transmission that assumes that sources begin constant rate transmissions of data at Poisson time points for random lengths of time. This model has been a popular one as analysts attempt to provide explanations for observed features in telecommunications data such as selfsimilarity, long range dependence and heavy tails. We survey some features of this model in cases where transmission length distributions have (a) tails so heavy that means are infinite, (b) heavy tails with finite mean and infinite variance and (c) finite variance. We survey the selfsimilarity properties of various descriptor processes in this model and then present analyses of four data sets which show that certain features of the model are consistent with the data while others are contradicted. The data sets are 1) the Boston University 1995 study of web sessions, 2) the UC Berkeley home IP HTTP data collected in November 1996, 3) tr...
System Support for Scalable and Fault Tolerant Internet Services
 IN IFIP INTERNATIONAL CONFERENCE ON DISTRIBUTED SYSTEMS PLATFORMS AND OPEN DISTRIBUTED PROCESSING (MIDDLEWARE '98), LAKE DISTRICT
, 1998
"... Over the past few years, we have seen the proliferation of Internetbased services ranging from search engines and map services to videoondemand servers. All of these kinds of services need to be able to provide guarantees of availability and scalability to their users. With millions of users on t ..."
Abstract

Cited by 15 (8 self)
 Add to MetaCart
Over the past few years, we have seen the proliferation of Internetbased services ranging from search engines and map services to videoondemand servers. All of these kinds of services need to be able to provide guarantees of availability and scalability to their users. With millions of users on the Internet today, these services must have the capacity to handle a large number of clients and remain available even in the face of extremely high load. In this paper, we present a generic architecture for supporting such Internet applications. We provide a substrate for Scalable Network Services (SNS), on top of which application developers can design their services without worrying about the details of service management. We back our design with three realworld services: a web distillation proxy, a proxybased webbrowser for PDAs, and an MBone archive server.
Stability and Performance Analysis of HIPERLAN
 Proc. IEEE INFOCOM ’98
, 1998
"... . This paper thoroughly analyses the HIPERLAN MAC protocol, which is a standard for Wireless Local Area Networks (WLANs) defined by the European Telecommunications Standards Institute (ETSI). Since the HIPERLAN MAC protocol belongs to the class of Carrier Sensing Multiple Access with Collision Avoid ..."
Abstract

Cited by 14 (0 self)
 Add to MetaCart
. This paper thoroughly analyses the HIPERLAN MAC protocol, which is a standard for Wireless Local Area Networks (WLANs) defined by the European Telecommunications Standards Institute (ETSI). Since the HIPERLAN MAC protocol belongs to the class of Carrier Sensing Multiple Access with Collision Avoidance (CSMA/CA), in the first part of the paper we analyze the HIPERLAN stability problem. In the second part we present an in depth performance analysis, by simulation, of the HIPERLAN MAC protocol. The analysis is performed by considering data traffic patterns (hereafter advanced data traffic) which have a very similar shape to traffic generated by WWW applications. Furthermore, in the analysis we consider Poissonian data traffic too. Our conclusion is that the HIPERLAN MAC protocol performs satisfactorily, although performance measures with advanced traffic are worse than the corresponding performance measures with Poissonian traffic. Furthermore, we broadened our analysis to include the i...
Benefits of packet aggregation in adhoc wireless network
, 2003
"... Emerging mesh networks build on IEEE 802.11 wireless local area networking technology to deliver Internet access to mobile users or to areas where traditional wired access technologies are unprofitable. Such deployments are highly bandwidth constrained; in dense urban environments many users are com ..."
Abstract

Cited by 14 (0 self)
 Add to MetaCart
Emerging mesh networks build on IEEE 802.11 wireless local area networking technology to deliver Internet access to mobile users or to areas where traditional wired access technologies are unprofitable. Such deployments are highly bandwidth constrained; in dense urban environments many users are competing for the wireless medium and longhaul wireless backbone connections have significant contention. In addition, IEEE 802.11 networks suffer from a high packet overhead—much higher than Ethernet, for example. This overhead consumes a significant share of the theoretical capacity. As networks advance to higher bitrates the proportion of this overhead increases, because packet preambles are transmitted at the lowest bitrate. We address this challenge through an adaptive, connectionagnostic aggregation mechanism that can reduce this overhead by combining multiple smaller packets into larger ones. One method uses a small delay on packets to enable the distributed aggregators to collect smaller packets from different network connections for assembling a larger packet. The second method automatically increases aggregation activity with network congestion; thus, it does not impose delays on network traffic in situations of low media utilization and also reduces delays under periods of high load. A prototype on a wireless testbed and simulation results for larger community network deployments show significant improvements in actual network capacity, ranging from a 60270 % improvement for traditional “adhoc” scenarios and to a 100500 % improvement for wireless networks using unidirectional antennas.