Performance Analysis Of Computer Systems Research Articles

Report on the Second International Workshop on Teaching Performance Analysis of Computer Systems 2023

Report on the Second International Workshop on Teaching Performance Analysis of Computer Systems 2023

Performance analysis of computer systems with Weibull distribution subject to software upgrade and load recovery

Performance analysis of computer systems with Weibull distribution subject to software upgrade and load recovery

On Quantum Algorithms for Random Walks in the Nonnegative Quarter Plane

It is well known that strong connections exist between random walks (RWs) in the nonnegative quarter plane and the mathematical performance modeling, analysis and optimization of computer systems and communication networks. Examples include adaptive threshold-based load balancing in concert with affinity scheduling (e.g., [16, 10]), scheduling parallel jobs on parallel computer systems (e.g., [15]), and asymptotic scalabilty in wireless and linear loss networks (e.g., [9]).

Report on the First International Workshop on Teaching Performance Analysis of Computer Systems 2021

Report on the First International Workshop on Teaching Performance Analysis of Computer Systems 2021

CostEfficient Dynamic Management of Cloud Resources through Supervised Learning

Brief Biography: Muhammad Wajahat is a PhD candi- date in the department of Computer Science at Stony Brook University. He works in the Performance Analysis of Com- puter Systems (PACE) Lab, under the supervision of Dr. Anshul Gandhi. Before joining PACE LAb, he graduated with distinction from Lahore University of Management Sci- ences (LUMS), Pakistan, with an honor thesis that investi- gated performance improvement of cloud gaming using Soft- ware De ned Networking (SDN). His current research fo- cuses on performance modeling and resource management to improving performance and resource utilization in cloud systems. He has worked with leading cloud research groups at IBM Research and AT&T Labs Research on application autoscaling and hotspot mitigation in private clouds.

APEM — Approximate Performance Evaluation for Multi-Core Computers

Mean Value Analysis (MVA) has long been a standard approach for performance analysis of computer systems. While the exact load-dependent MVA algorithm is an efficient technique for computer system performance modeling, it fails to address multi-core computer systems with Dynamic Frequency Scaling (DFS). In addition, the load-dependent MVA algorithm suffers from numerical difficulties under heavy load conditions. The goal of our paper is to find an efficient and robust method which is easy to use in practice and is also accurate for performance prediction for multi-core platforms. The proposed method, called Approximate Performance Evaluation for Multi-core computers (APEM), uses a flow-equivalent performance model designed specifically to address multi-core computer systems and identify the influence on the CPU demand of the effect of DFS. We adopt an approximation technique to estimate resource demands to parametrize MVA algorithms. To validate the application of our method, we investigate three case studies with extended TPC-W benchmark kits, showing that our method achieves better accuracy compared with other commonly used MVA algorithms. We compare the three different performance models, and we also extend our approach to multi-class models.

Probability and statistics with reliability, queueing, and computer science applications (trivedi, k.s.) [book review

Engineers, computer scientists, and networking engineers require robust techniques and methods to analyze algorithms, computer systems, protocol behavior, routing algorithms, and congestion in networks. A perfect source of the related information is provided with this second edition of a classic book. As we have not commented on this book before in our magazine, we fill this gap now, taking the opportunity of the situation that Wiley decided to publish the paperback version. The book provides a comprehensive introduction to probability, stochastic processes and statistics, and illustrates how these concepts can be applied in analysis, modeling, and evaluation of the reliability and performance of computer and communication systems. This is a very useful book for students, as well as reliability and computer science engineering professionals. It is focused on the performance, dependability and performability analysis of computer systems. The main strength of this book is the vast number of examples and exercises, which serve to illustrate the complex theoretical concepts and help the reader to a better understanding of the important topics. I highly recommend this book for academic use as a textbook and for researchers and professionals as a useful reference.

Experimental analysis of the effect of filtering perceptions in BDI agents

One of the BDI paradigm's major concerns is the lack of control over the agents' perceptions. Without having any form of goal-directed perceptions, agents may be flooded by irrelevant information thus causing an unjustified increase in processing time. This issue becomes critical when one needs to develop agents to be integrated with virtual environments or simulators, or even in the case of embedded agents, as robots. In order to provide greater control on the agents' perceptions and to reduce their time response, this work proposes to incorporate a filtering perception mechanism within the Jason interpreter, aiming to eliminate irrelevant perceptions in order to reduce processing time. To this end, some types of pre-defined filters proposed in the literature have been implemented and their effect experimentally evaluated in three different simulated and one embedded agent experiments. Using a full factorial experiment design, a technique used in computer systems performance analysis, it was shown that applying perception filters can reduce up to 80% of an agent's processing time, without significantly affecting its performance measured in terms of its utility function.

Modelling and analysis of gene regulatory networks based on the G-network

G-networks are a class of stochastic models that have had a broad range of applications ranging from the performance analysis of computer systems and networks to the modelling of gene regulatory networks. Gene regulatory networks consist of thousands of genes and proteins which are dynamically interacting with each other. Once these regulatory structures are revealed, it is necessary to understand their dynamical behaviours since pathway activities could be changed by their given conditions. This review mainly focuses on a stochastic GRN modelling techniques based on G-networks which provide the analytical steady-state solution of a system for efficient GRN dynamics modelling. Three applications of the G-network model to GRNs show that this novel approach can serve to detect abnormalities from protein expression data, and that they can help to explicit the behaviour of complicated GRN models by dividing the gene regulatory processes into DNA and protein layers.

An Efficient Use of Principal Component Analysis in Workload Characterization-A Study

PCA is a useful statistical technique that has found application in fields such as face recognition, image compression, dimensionality reduction, Computer System performance analysis etc. It is a common technique for finding patterns in data of high dimension. In this paper, we present the basic idea of principal component analysis as a general approach that extends to various popular data analysis techniques. We state the mathematical theory behind PCA and focus on monitoring system performance using the PCA algorithm. Next, an Eigen value-Eigenvector dynamics is elaborated which aims to reduce the computational cost of the experiment. The Mathematical theory is explored and validated. For the purpose of illustration we present the algorithmic implementation details and numerical examples over real time and synthetic datasets.

Time Domain Measurement Representation in Computer System Diagnostics and Performance Analysis

Time analysis is a common approach for testing and detecting methods for the performance analysis of computer systems. In the article it is shown, that measuring and identifying performances based on a benchmark is not sucient for the proper analysis of the computer systems behavior. The response time of the process is often composed of the execution of many subprocesses or many paths of execution. Under this assumption, it is presented, that both convolution and deconvolution methods can be helpful in obtaining time distributions and modeling of complex processes. In such a modeling the analysis of measurement errors is very important and was taken into consideration. The example of using the methods in buering process is also discussed.

Perspectives on network calculus

ACM Sigcomm 2006 published a paper [26] which was perceived to unify the deterministic and stochastic branches of the network calculus (abbreviated throughout as DNC and SNC) [39]. Unfortunately, this seemingly fundamental unification---which has raised the hope of a straightforward transfer of all results from DNC to SNC---is invalid. To substantiate this claim, we demonstrate that for the class of stationary and ergodic processes, which is prevalent in traffic modelling, the probabilistic arrival model from [26] is quasi-deterministic, i.e., the underlying probabilities are either zero or one. Thus, the probabilistic framework from [26] is unable to account for statistical multiplexing gain, which is in fact the raison d'être of packet-switched networks. Other previous formulations of SNC can capture statistical multiplexing gain, yet require additional assumptions [12], [22] or are more involved [14], [9] [28], and do not allow for a straightforward transfer of results from DNC. So, in essence, there is no free lunch in this endeavor. Our intention in this paper is to go beyond presenting a negative result by providing a comprehensive perspective on network calculus. To that end, we attempt to illustrate the fundamental concepts and features of network calculus in a systematic way, and also to rigorously clarify some key facts as well as misconceptions. We touch in particular on the relationship between linear systems, classical queueing theory, and network calculus, and on the lingering issue of tightness of network calculus bounds. We give a rigorous result illustrating that the statistical multiplexing gain scales as Ω(√ N ), as long as some small violations of system performance constraints are tolerable. This demonstrates that the network calculus can capture actual system behavior tightly when applied carefully. Thus, we positively conclude that it still holds promise as a valuable systematic methodology for the performance analysis of computer and communication systems, though the unification of DNC and SNC remains an open, yet quite elusive task.

Distributed Response Time Analysis of GSPN Models with MapReduce

Generalized stochastic Petri nets (GSPNs) are widely used in the performance analysis of computer and communications systems. Response time densities and quantiles are often key outputs of such analysis. These can be extracted from a GSPN’s underlying semi-Markov process using a method based on numerical Laplace transform inversion. This method typically requires the solution of thousands of systems of complex linear equations, each of rank n, where n is the number of states in the model. For large models substantial processing power is needed and the computation must therefore be distributed. In this paper we describe the implementation of a response time analysis module for the Platform Independent Petri net Editor (PIPE2) which interfaces with Hadoop, an open-source implementation of Google’s MapReduce distributed programming environment, to provide distributed calculation of response time densities in GSPN models. The software is validated with results calculated analytically as well as simulated results for larger models. Excellent scalability is shown.

A Formal Framework for Stochastic Discrete Event System Specification Modeling and Simulation

We introduce an extension of the classic Discrete Event System Specification (DEVS) formalism that includes stochastic features. Based on the use of the probability spaces theory we define the stochastic DEVS (STDEVS) specification, which provides a formal framework for modeling and simulation of general non-deterministic discrete event systems. The main theoretical properties of the STDEVS framework are treated, including a new definition of legitimacy of models in the stochastic context and a proof of STDEVS closure under coupling. We also illustrate the new stochastic modeling capabilities introduced by STDEVS and their relation with those found in classic DEVS. Practical simulation examples are given involving performance analysis of computer systems and hybrid modeling of networked control systems, applications where the modeling of stochastic components is vital.

Bio-PEPA: A framework for the modelling and analysis of biological systems

In this work we present Bio-PEPA, a process algebra for the modelling and the analysis of biochemical networks. It is a modification of PEPA, originally defined for the performance analysis of computer systems, in order to handle some features of biological models, such as stoichiometry and the use of general kinetic laws. Bio-PEPA may be seen as an intermediate, formal, compositional representation of biological systems, on which different kinds of analyses can be carried out. Bio-PEPA is enriched with some notions of equivalence. Specifically, the isomorphism and strong bisimulation for PEPA have been considered and extended to our language. Finally, we show the translation of a biological model into the new language and we report some analysis results.

Guest Editors' Introduction to the Special Issue on Quantitative Evaluation of Computer Systems

OFTWARE systems are expected to meet a multitude of quantitative constraints, such as timely response, performability, and efficient use of energy and other resources. Quantitative evaluation of systems refers to a set of techniques which can be applied to analyze such systems and obtain a range of quantitative characteristics of their performance, for example, the probability of satisfying a request within a given time or the appropriate number of replicas in a quorum air traffic control system. In this special issue, we publish a diverse range of novel research work which contributes to the development of quantitative evaluation of computer systems from a number of perspectives. These include theoretical foundations, models and modeling languages, analysis methods, simulation, verification, and software tools. Queueing networks have long been used for performance analysis of computer systems. This issue includes two papers related to queueing networks. In the first, “Enhanced Modeling and Solution of Layered Queueing Networks,” the layered queueing networks modeling formalism is presented. This modeling language allows a structured, hierarchical queueing network to be built up representing the resource usage and contention within a system. Analysis techniques which make use of the hierarchical structure allow complex models to be analyzed to address questions about the performance and dependability of the system. The second queueing paper, “CoMoM: Efficient Class-Oriented Evaluation of Multiclass Performance Models,” describes an improved solution technique for multiclass queueing network models. For more than 25 years, mean value analysis (MVA) has been one of the most frequently applied analysis methods of these models. This paper proposes an efficient rearrangement of recursive computational steps (referred to as the CoMoM method) such that it minimizes the number of recursive steps needed to solve the model. The CoMoM method makes possible the approximate analysis of much larger queueing network models, with tens of classes and thousands of nodes. Another well-established quantitative modeling technique is based on timed and stochastic extensions of Petri nets. In “State-Density Functions over DBM Domains in the Analysis of Non-Markovian Models,” a solution technique for such a Petri net extension is considered. The analysis of stochastic Petri nets with exponentially distributed firing times is well-developed and widely applied in software systemsmodeling.Incontrast,thecaseofstochasticPetrinets

Workload Characterization an Essential Step in Computer Systems Performance Analysis - Methodology and Tools

Workload Characterization an Essential Step in Computer Systems Performance Analysis - Methodology and Tools

Modelling of Biochemical Reactions by Stochastic Automata Networks

This paper presents a stochastic modelling framework based on stochastic automata networks (SANs) for the analysis of complex biochemical reaction networks. Our approach takes into account the discrete character of quantities of components (i.e. the individual populations of the involved chemical species) and the inherent probabilistic nature of microscopic molecular collisions. Moreover, as for process calculi that have recently been applied to systems in biology, the SAN approach has the advantage of a modular design process being adequate for abstraction purposes. The associated composition operator leads to an elegant and compact representation of the underlying continuous-time Markov chain in form of a Kronecker product. SANs have been extensively used in performance analysis of computer systems and a large variety of numerical and simulative analysis algorithms exist. We illustrate that describing a biochemical reaction network by means of a SAN offers promising opportunities to get insight into the quantitative behaviour of systems in biology while taking advantage of the benefits of a compositional modelling approach.

Constrained Multivariate Extrapolation Models With Application to Computer Cache Rates

In this article we propose an approach to building multivariate regression models for prediction beyond the range of the data. The extrapolation model attempts to accurately estimate the high-level trend of the data, which can be extended in a natural way. The constraints of monotonicity and convexity/concavity play an important role in restricting the choice of the high level-trend, which otherwise would remain rather arbitrary. Our extrapolation model incorporates these constraints in multiple dimensions. We describe the trend as a nonnegative linear combination of twice-integrated multivariate B-splines and their variations. The specific basis functions in our approach are chosen so that any such combination is a plausible a priori model. As a result, basis function coefficients can be optimized to best fit the data without losing control over the high-level trend of the extrapolation model. Our approach also allows the use of standard model selection techniques. We illustrate this by applying cross-validation to roughness penalty parameter selection. We demonstrate the validity of our approach by successfully applying it to modeling computer cache miss rates—a key problem in computer system performance analysis.

Smooth Isotonic Additive Interaction Models With Application to Computer System Architecture Design

The regression relationship between a predictor vector and a response is assumed to be smooth, additive with low-order interactions, and monotone in some of the predictors. We consider two modeling approaches. The first approach is to fit an unconstrained additive model first and subsequently impose the isotonic constraint to cause the least perturbation of the fit. The second approach optimizes the fitting criterion that includes a roughness penalty over isotonic smoothing splines. The advantage of the first approach is its modularity and ease of implementation. One can incorporate a variety of smoothing methods as building blocks of the unconstrained model. In the article we develop the theory for this approach. In particular, we show that one can use a combination of off-the-shelf smoothing, fitting, and isotonic routines. However, it is empirically shown that fitting an isotonic smoothing spline model produces superior results over a broad range of target functions. The methodology is applied to modeling computer cache miss rates, a key problem in computer system performance analysis.