paper
Oufimtsev, A. and Murphy, L. Method Input Parameters and Performance of EJB Applications. Studia Informatica Universalis Journal, Regular Issue Vol.4 No. 1 pp. 41-52
Abstract
We investigated the impact of method input parameters on component performance, which is usually neglected during an application design stage. We evaluated a set of Commercial Off-The-Shelf (COTS) components which use Enterprise Java Beans (EJB) technology. These components were deployed on WebSphere Application Server and tested using a custom-built remote client. The client keeps track of execution times, while the server has JProbe Profiler embedded in the EJB container to monitor the interactions. For test purposes we used only stateless session beans, allowing us to concentrate on the possible dependency of server performance on input parameter variations. Test results show that significant performance impacts can be caused by a relatively small number of input parameter dependencies.
Jing Xu, Alexandre Oufimtsev, Murray Woodside, Liam Murphy,Performance Modeling and Prediction of Enterprise JavaBeans with Layered Queuing Network Templates, SAVCBS'05 workshop, ACM
Abstract
Component technologies, such as Enterprise Java Beans (EJB) and .NET, are used in enterprise servers with requirements for high performance and scalability. This work considers performance prediction from the design of an EJB system, based
on the modular structure of an application server and the application components. It uses layered queueing models, which
are naturally structured around the software components. This paper describes a framework for constructing such models, based on layered queue templates for EJBs, and for their inclusion in the server. The resulting model is calibrated and validated by comparison with an actual system.
Ufimtsev, A., Murphy, L., Kucharenka A, Impact of Virtual Memory Managers on Performance of J2EE Applications In Proceedings of CBSE conference, June. 2006
Abstract.
We investigate the impact of Operating System's Virtual Memory Managers (VMMs) on performance of enterprise applications.
By taking various popular branches of the Linux kernel and modifying their VMM settings, one can see the effects it introduces on ECPerf J2EE Benchmark. JBoss application server is used to run ECPerf. Our tests show that even the change of one parameter in VMM can have signicant performance impacts. Performance of various kernel branches is compared. Parameter sensitivity and infuence of specific settings arepresented.
A. Ufimtsev, T. Parsons, L. Patcas, J. Murphy and L. Murphy, Introducing Performance Engineering by means of Tools and Practical Exercises, accepted for presentation at CASCON 2006: DUBLIN SYMPOSIUM, In association with CASCON 2006: 16th Annual International Conference on Computer Science and Software Engineering, IBM Dublin Center for Advanced Studies, Ireland, October 16-17, 2006.
Abstract
Many software engineers complete their education without an introduction to the most basic performance engineering concepts. IT specialists need to be educated with a basic degree of performance engineering knowledge, so they are aware of why and how certain design and development decisions can lead to poor performance of the resulting software systems. To help address this need, the School of Computer Science and Informatics at University College Dublin offered a final year undergraduate/first year postgraduate module on ”Performance of Computer Systems” in Autumn 2005. In this
A. Ufimtsev and L. Murphy, Performance Modelling of a JavaEE Component Application using Layered Queuing Networks: Revised Approach and a Case Study, accepted for presentation at the Specification and Verification of Component-Based Systems (SAVCBS'06) Workshop at ACM SIGSOFT 2006/FSE-14, Portland, OR, USA, November 10-11, 2006
Abstract:
Nowadays component technologies are an integral part of any enterprise production environment. Performance and
scalability are among the key properties of such systems. Using Layered Queuing Networks (LQN), one can predict the
Ufimtsev, A. and Murphy, L. Automatic System for Linux Kernel Performance Testing, In Proceedings of Linux Symposium 2006, Vol 2, pp 403-408, July 2006, Ottawa
Abstract:
Comprehensive automatic kernel tests help to prevent instability issues and performance regressions, while lack thereof is considered to be a significant contributor to the kernel quality problem. Most of the available kernel testing projects utilize 'big iron' machines in the hope of detecting potential problems in the kernel tree. Though useful, the test results produced by these projects are quite unrepresentative, since they tend to test kernel performance on very specific hardware with very specific configurations. However, a kernel can run on all kinds of platforms supporting a variety of hardware, and has a huge number of configurable parameters. We argue that only by opening a test system to the community and aggregating the results from a variety of sources can one get a comprehensive picture of the kernel's performance status. The goal of our project is an automatic testing system that downloads, compiles, installs, and initiates performance tests of kernel branch snapshots. The resulting data are sent to a server and then made available to developers via a web interface. If regressions are found, our system can pinpoint the problem down to submitter. The client part of the system performs 'jit' tree checkouts, compilations, installations, and standardized performance tests. Focusing on successful automation and configuration of the testing process, our client is available as a package, thus making it easy to install and configure. The centralized server software manages data submissions, data aggregation, and results presentation. The emphasis of the server part is placed on regression detection and accidental/malicious input elimination. A beta-test of the system is performed with clients running on variety of platforms, including x86, alpha, and PPC.
|
