We present the Java Modelling Tools (JMT) suite, an integrated framework of tools for performance evaluation computer systems using queueing models. The suite offers a rich user interface that simplifies definition models by means wizard dialogs and graphical design workspace. features JMT span wide range state-of-the-art methodologies including discrete-event simulation, mean value analysis product-form networks, analytical identification bottleneck resources in multiclass environments,...

The performance of a system is determined by its characteristics as well the composition load being processed. Hence, quantitative description fundamental part all evaluation studies. Several methodologies for construction workload models, which are functions objective study, architecture to be analyzed, and techniques adopted, presented. A survey few applications these various types systems (i.e., batch, interactive, database, network-based, parallel, supercomputer), given.< <ETX...

The concept of a processor working set (PWS) as single value parameter for characterizing the parallel program behavior is introduced. Through detailed experimental studies different algorithms on transputer-based multiprocessor machine, it shown that PWS robust measure workload system. It allocation strategies based provide significantly better throughput-delay characteristics. robustness further demonstrated by showing policies allocate processors more than are inferior in performance to...

This paper presents a simulation analysis of the impact mobility on TCP/IP augmented with features to support host in wide area networks. Our results show that existing version TCP can yield low throughput highly mobile environments due fact cannot discriminate packets dropped hand-offs those congestion one or more network resources. As result, invokes recovery process when are lost during internetwork host. We investigate proposal which transport layer explicitly receives information from...

In this paper we systematically examine various performance issues involved in the coordinated allocation of processor and disk resources large-scale parallel computer systems. Models are formulated to investigate I/O computation behavior programs workloads, analyze scheduling policies under such workloads. These models parameterized by measurements programs, they solved via analytic methods simulation. Our results provide important insights into applications resource management strategies...

This work summarizes our research on the topic of application unsupervised learning algorithms to problem intrusion detection, and in particular main results network detection. We proposed a novel, two tier architecture for capable clustering packet payloads correlating anomalies stream. show experiments we conducted such architecture, give performance results, compare achievements with other comparable existing systems.

Java Modelling Tools (JMT) is a suite of open source applications for performance evaluation and workload characterization computer communication systems based on queueing networks. JMT includes tools (JWAT), solution networks with analytical algorithms (JMVA), simulation general-purpose models (JSIM), bottleneck identification (JABA), teaching support Markov chain underlying (JMCH). This tutorial summarizes the main features that compose suite. Furthermore, using composite case study, we...

This paper describes JSIM: the simulation module of Java Modelling Tools (JMT), an open-source fully-portable suite for capacity planning studies. The simulator has been purposely developed to help both unexperienced and advanced users. Most difficult decisions that are needed in order run simulations properly, such as detection transient part samples be discarded, have automated. tool also provides guidance over graphical design network analysis plot results. What-if parametric analyses...

Distributed application systems are composed of classes objects with instances that interact to accomplish common goals. Such can have many users types requests. Furthermore, the relative load these shift throughout day, causing changes system behavior and bottlenecks. When designing deploying such systems, it is necessary determine a process replication threading policy for server processes contain objects, as well activation policies. To avoid bottlenecks, must support all possible...

A fundamental understanding of the interplay between computation and I/O activities in parallel applications that manipulate huge amounts data is critical to achieving good application performance, as well correctly characterizing workloads large-scale high-performance systems. We present a formal model behavior CPU interactions scientific applications, from which we derive various formulas characterize performance. Our captures activity at different levels granularity, where results are...

Queuing network models of modern computing systems must consider a large number components (e.g., Web servers, DB application firewall, routers, networks) and hundreds customers with very different resource requirements. The complexity such makes the exact solution techniques prohibitively expensive, motivating research on approximate methods. This paper proposes an interpolation-matching framework that allows unified view for closed product-form queuing networks. Depending upon...

The load generated by new types of communications services related to multimedia and video transmission is becoming one the major sources traffic in WAN networks. Modeling this type a prerequisite for any performance study. We approach load-characterization problem from global point view analyzing set 20 streams. developed resource-, subject-, scene-oriented characterizations coded have also implemented multidimensional data-analysis techniques applied "scene working set" approach, as well...

It is known that the resources limit overall performance of a system are congested ones, referred to as bottlenecks. From knowledge bottleneck stations, it possible, with limited computational effort, derive asymptotic values several indices. While identifying stations well-established practice under single-class workload, no simple methodology for multiclass models exists. We present new algorithms bottlenecks in queueing networks constant-rate servers. show how application assessed...

We propose the Geometric Bounds (GBs), a new family of fast and accurate noniterative bounds on closed queueing network performance metrics that can be used in online optimization distributed applications. Compared to state-of-the-art techniques such as Balanced Job (BJBs), GB achieves higher accuracy at similar computational costs, limiting worst- case bounding error typically within 5-13 percent when, for BJB, it is usually range 15-35 percent. Optimization problems are solved with GBs...

Flexibility is an important characteristic of cloud computing: users can dynamically lease resources only when needed. While this feature brings along various benefits, there are several drawbacks and among them performance unpredictability a major issue. In paper we investigate the effects new source variability introduced by flexible CPUs allocation policy supported IaaS providers. According to policy, with different computing power may be allocated same VM type. We will show through...

NoSQL databases such as Apache Cassandra have attracted large interest in recent years thanks to their high availability, scalability, flexibility and low latency. Still, there is limited research work on performance engineering methods for databases, which yet are needed since these systems a

In this paper, processor scheduling policies that "save" processors are introduced and studied. a multiprogrammed parallel system, "processor saving" policy purposefully keeps some of the available idle in presence work to be done. The conditions under which saving can more effective than their greedy counterparts, i.e., never leave done, examined. Sensitivity analysis is performed with respect application speedup, system size, coefficient variation applications' execution time, variability...