Results 1  10
of
81
A Toolset for Performance Engineering and Software Design of ClientServer Systems
 Performance Evaluation
, 1996
"... TimeBench/SRVN is a prototype toolset for computeraided design and performance analysis of software, with an emphasis on distributed clientserver systems. The performance behaviour of such systems may defy intuition because it involves factors in the software design (such as the partitioning of th ..."
Abstract

Cited by 33 (13 self)
 Add to MetaCart
(Show Context)
TimeBench/SRVN is a prototype toolset for computeraided design and performance analysis of software, with an emphasis on distributed clientserver systems. The performance behaviour of such systems may defy intuition because it involves factors in the software design (such as the partitioning of the functionality and the frequency with which requests will be made to each server) and in the configuration of the distributed system (including replication of services, the distribution of data, and the speed of network access). The novelty of the tool consists in providing support both for developing design specifications and also for performance analysis. The integrated approach avoids the semantic gap between a designer's domain and the performance modeling domain, and assists the designer to explore factors that impact the performance of a design. The performance models are based on the Stochastic Rendezvous Network (SRVN) formalism for clientserver systems with synchronous service req...
The Automatic Generation of Software Performance Models From a Prototype
 In International Workshop on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (MASCOTS'95
, 1995
"... Early performance estimates for a new software system aid the design process by providing feedback when design decisions can be easily revised. Unfortunately, constructing a performance model of a distributed and concurrent software system can require significant effort. An automated performance mod ..."
Abstract

Cited by 30 (6 self)
 Add to MetaCart
Early performance estimates for a new software system aid the design process by providing feedback when design decisions can be easily revised. Unfortunately, constructing a performance model of a distributed and concurrent software system can require significant effort. An automated performance model generation technique is described that reduces the model building effort by providing: easy specification of performance experiments, empirical estimates for model parameters, automated model generation, and support for different types of models. A prototype is used to describe a software system, from which causal traces (angio traces) are recorded during execution. These traces are then processed into sequences of resource demands (workthreads), aggregated into system execution descriptions (workthread classes), and combined to generate a performance model. The technique can also be applied at other stages of the development process, including the redesign of existing software. Page ii...
Symbolic Performance Modeling of Parallel Systems
 In Proc. of IEEE Transactions on Parallel and Distributed Systems
, 2003
"... Abstract—Performance prediction is an important engineering tool that provides valuable feedback on design choices in program synthesis and machine architecture development. We present an analytic performance modeling approach aimed to minimize prediction cost, while providing a prediction accuracy ..."
Abstract

Cited by 24 (0 self)
 Add to MetaCart
Abstract—Performance prediction is an important engineering tool that provides valuable feedback on design choices in program synthesis and machine architecture development. We present an analytic performance modeling approach aimed to minimize prediction cost, while providing a prediction accuracy that is sufficient to enable major code and data mapping decisions. Our approach is based on a performance simulation language called PAMELA. Apart from simulation, PAMELA features a symbolic analysis technique that enables PAMELA models to be compiled into symbolic performance models that trade prediction accuracy for the lowest possible solution cost. We demonstrate our approach through a large number of theoretical and practical modeling case studies, including six parallel programs and two distributedmemory machines. The average prediction error of our approach is less than 10 percent, while the average worstcase error is limited to 50 percent. It is shown that this accuracy is sufficient to correctly select the best coding or partitioning strategy. For programs expressed in a highlevel, structured programming model, such as dataparallel programs, symbolic performance modeling can be entirely automated. We report on experiments with a PAMELA model generator built within a dataparallel compiler for distributedmemory machines. Our results show that with negligible program annotation, symbolic performance models are automatically compiled in seconds, while their solution cost is in the order of milliseconds. Index Terms—Performance prediction, parallel processing, analytic performance modeling. æ 1
TwoLevel Iterative Queuing Modeling of Software Contention
, 2002
"... Being able to model contention for software resources (e.g., a critical section or database lock) is paramount to building performance models that capture all aspects of the delay encountered by a process as it executes. Several methods have been offered for dealing with software contention and with ..."
Abstract

Cited by 18 (3 self)
 Add to MetaCart
Being able to model contention for software resources (e.g., a critical section or database lock) is paramount to building performance models that capture all aspects of the delay encountered by a process as it executes. Several methods have been offered for dealing with software contention and with message blocking in clientserver systems. This paper presents a general, straightforward, easy to understand and implement, approach to modeling software contention using queuing networks. The approach, called SQNHQN, consists of a twolevel iterative process. Two queuing networks are considered: one represents software resources (SQN) and the other hardware resources (HQN). Multiclass models are allowed and any solution techniqueexact or approximatecan be used at any of the levels. This technique falls in the general category of fixedpoint approximate models and is similar in nature to other approaches. The main difference lies in its simplicity. The process converges very fast in the examples examined. The results were validated against global balance equation solutions and are very accurate.
A Tree Convolution Algorithm for the Solution of Queueing Networks
 Commun. ACM
, 1982
"... A new algorithm called the tree convolution algorithm for the computation of normalization constants and performance measures of productform queueing algorithms, in the solution of networks with many service centers and many sparse routing chains. (A network is said to have sparse routing chains ..."
Abstract

Cited by 17 (2 self)
 Add to MetaCart
A new algorithm called the tree convolution algorithm for the computation of normalization constants and performance measures of productform queueing algorithms, in the solution of networks with many service centers and many sparse routing chains. (A network is said to have sparse routing chains if the chains visit, on the average, only a small fraction of all centers in the net work.) In such a network, substantial time and space savings can be achieved by exploiting the network's routing information. The time and space reductions are made possible by two features of the algorithm: (1) the sequence of array convolutions to compute a normalization constant is determined according to the traversal of a tree; and (2) the convolutions are performed between arrays that are smaller than arrays used by existing algorithms. The routing information of a given network is used to configure the tree to reduce the algorithms time and space requirements; some effective heuristics for optimization are de scribed herein. An exact solution of a communication network model with 64 queues and 32 routing chains is illustrated.
On Performance Prediction of Parallel Computations with Precedent Constraints
, 1994
"... Performance analysis of concurrent executions in parallel systems has been recognized as a challenging problem. The aim of this research is to study approximate but ecient solution techniques for this problem. We model the structure of a parallel machine and the structure of the jobs executing on ..."
Abstract

Cited by 15 (0 self)
 Add to MetaCart
Performance analysis of concurrent executions in parallel systems has been recognized as a challenging problem. The aim of this research is to study approximate but ecient solution techniques for this problem. We model the structure of a parallel machine and the structure of the jobs executing on such a system. We investigate rich classes of jobs, which can be expressed by series, paralleland, parallelor, and probabilisticfork. We propose an efficient performance prediction method for these classes of jobs running on a parallel environment which is modeled by a standard queueing network model. The proposed prediction method is computationally efficient, it has polynomial complexity in both time and space. The time complexity is O(C²N²K) and the space complexity is O(C²N²K), where C is the number of job classes in the system, the number of tasks in each job class is O(N), and K is the number of service centers in the queueing model. The accuracy of the approxi...
Bottlenecks Identification in Multiclass Queueing Networks using Convex Polytopes
, 2004
"... It is known that the resources that limit the overall performance of the system are the congested ones, referred to as bottlenecks. From the knowledge of the bottleneck stations with a limited computational effort it is possible to derive asymptotic values of several performance indices. While ident ..."
Abstract

Cited by 15 (3 self)
 Add to MetaCart
It is known that the resources that limit the overall performance of the system are the congested ones, referred to as bottlenecks. From the knowledge of the bottleneck stations with a limited computational effort it is possible to derive asymptotic values of several performance indices. While identifying the bottleneck stations under a singleclass workload is a wellestablished practice, no simple methodology for multiclass models exists. In this paper we present new algorithms for identifying the bottlenecks in multiclass queueing networks with constantrate servers. We show how the application of assessed techniques, such as the ones related to the convex polytopes, can provide insights on the performance of a queueing network. The application of our techniques to the asymptotic analysis of closed productform networks is also investigated.
An Efficient Algorithm for the Exact Analysis of Multiclass Queueing Networks with Large Population Sizes
, 2006
"... We introduce an efficient algorithm for the exact analysis of closed multiclass productform queueing network models with large population sizes. We adopt a novel approach, based on linear systems of equations, which significantly reduces the cost of computing normalizing constants. With the propose ..."
Abstract

Cited by 14 (9 self)
 Add to MetaCart
We introduce an efficient algorithm for the exact analysis of closed multiclass productform queueing network models with large population sizes. We adopt a novel approach, based on linear systems of equations, which significantly reduces the cost of computing normalizing constants. With the proposed algorithm, the analysis of a model with N circulating jobs of multiple classes requires essentially the solution of N linear systems with order independent of population sizes. A distinguishing feature of our approach is that we can immediately apply theorems, solution techniques, and decompositions for linear systems to queueing network analysis. Following this idea, we propose a block triangular form of the linear system that further reduces the requirements, in terms of both time and storage, of an exact analysis. An example illustrates the efficiency of the resulting algorithm in presence of large populations.
Mean Value Analysis for Queueing Network Models with Intervals as Input Parameters
, 1998
"... Mean value analysis (MVA) is a wellknown solution technique for separable closed queueing networks used in performance modeling of computer and communication systems. In many cases, like for sensitivity analysis or with inaccurate model input parameters, intervals are more appropriate as model inpu ..."
Abstract

Cited by 14 (12 self)
 Add to MetaCart
(Show Context)
Mean value analysis (MVA) is a wellknown solution technique for separable closed queueing networks used in performance modeling of computer and communication systems. In many cases, like for sensitivity analysis or with inaccurate model input parameters, intervals are more appropriate as model inputs than single values. This paper presents a version of the MVA algorithm for separable closed queueing networks with one customer class consisting of loadindependent queueing centers as well as delay devices, which accepts both single values and intervals as input parameters in arbitrary combination. Monotonicity of the model outputs with respect to all input parameters is proved and these monotonicity properties are used to construct a low cost intervalversion of the MVA algorithm providing exact output intervals as results. Thus, dependency problems commonly arising with the interval evaluation of arithmetic expressions are avoided without significant increase in computation costs. Addit...
Proportional fairness and its relationship with multiclass queueing networks
, 2009
"... We consider multiclass singleserver queueing networks that have a product form stationary distribution. A new limit result proves a sequence of such networks converges weakly to a stochastic flow level model. The stochastic flow level model found is insensitive. A large deviation principle for the ..."
Abstract

Cited by 14 (7 self)
 Add to MetaCart
(Show Context)
We consider multiclass singleserver queueing networks that have a product form stationary distribution. A new limit result proves a sequence of such networks converges weakly to a stochastic flow level model. The stochastic flow level model found is insensitive. A large deviation principle for the stationary distribution of these multiclass queueing networks is also found. Its rate function has a dual form that coincides with proportional fairness. We then give the first rigorous proof that the stationary throughput of a multiclass singleserver queueing network converges to a proportionally fair allocation. This work combines classical queueing networks with more recent work on stochastic flow level models and proportional fairness. One could view these seemingly different models as the same system described at different levels of granularity: a microscopic, queueing level description; a macroscopic, flow level description and a teleological, optimization description. 1. Introduction. In