Breakfast, Registration

9:00 – 9:15

Welcome and Announcements

9:15 – 10:15

Moving Beyond Test and Guess - Using Modeling with Load Testing to Improve Web Application Readiness

This paper will discuss and demonstrate how performance modeling compliments load testing to improve the readiness of an application before deployment. After a discussion of the benefits and challenges of modeling, a case study of an actual load testing and modeling effort is presented. The case study walks through the entire model construction process, from data collection, through model construction, and model validation. The case study concludes with a description of how the model was used to guide and improve the load testing effort.

Richard Gimarc, HyPerformix

Richard Gimarc has over twenty-five years of technology experience in computer performance analysis and software development. Mr. Gimarc has broad experience in solving a variety of complex technology and performance problems affecting a wide range of applications and computing platforms ranging from standard business applications, to real-time systems, to operating systems, and Web-based applications. Mr. Gimarc has presented numerous times at technical conferences and has authored several technology papers related to performance engineering. He is currently employed by HyPerformix in Austin, TX.

10:15 – 10:30

Break

10:30 – 11:30

Confidentiality implications for Web transaction measurement and monitoring

Balancing the need for effective web transaction performance management while meeting compliance and confidentiality regulations is a key challenges that organizations face. Privacy requirements and stringent regulations such as GLBA and HIPAA affect a broad spectrum of organizations. Can you effectively monitor web performance and comply with regulations when transactions include secure and confidential data? This session looks at approaches that satisfy security requirements while providing access to business relevant performance data.

Alistair Croll, Coradiant

Alistair Croll co founded Coradiant, a leader in real-user Web performance management, where he is responsible for product engineering and business strategy. Previously, he was a principal with Networkshop, an Internet infrastructure consultancy to major networking firms on performance and QoS. He also has worked with 3Com, Primary Access, and Eicon Technology. He has managed and consulted on technologies including VPNs, QOS, access concentration, high availability, server performance, and collocated security. He is coauthor of Managing Bandwidth: Deploying QOS Across Enterprise Networks (Prentice Hall PTR, 1999) and has been active as a member of the Networld+Interop program committee. He was graduated from Dalhousie University with a B.Com (Hon.) in strategic marketing

11:30 – 12:15

Lunch

12:15 – 1:15

Application Performance Has Significantly Degraded. It Must Be the I/O Subsystem. Please Upgrade It!

This paper presents an I/O analysis undertaken due to a commonly reported performance problem of poor user or application response time. In this real-life case, users were experiencing poor response and the initial perspective was to blame the I/O subsystem. The conclusions from the analysis eliminated the I/O subsystem as the application performance bottleneck. This resulted in a win-win situation, saving both companies significant dollars from revamping an I/O infrastructure, which would have made little difference to application performance and properly focused attention on the true problem. This paper includes a review of the tools used in the analysis including EMC’s Workload Analyzer. This type of analysis can be applied to all I/O subsystem infrastructures including open systems and mainframe environments.

Alan Phelan, (i)Structure

An Irishman working in the IT data center field for 14 years. Previously for the last three years the manager for Storage Services at (i)Structure but now recently appointed the Director for UNIX and Windows services at (i)Structure.

1:15 – 2:15

Multiprocessor scalability in Microsoft Windows Server 2003

This presentation provides an overview of the multiprocessing support in the Microsoft Windows 2003 operating system, with an emphasis on scalability and other capacity planning issues. It also discusses specific features of the 32-bit and 64-bit processor architectures that provide the hardware basis for large scale multiprocessing systems. As a shared memory multiprocessing implementation, Windows Server 2003 is predictably vulnerable to saturation on the shared memory bus. The performance characteristics of cache coherent NUMA architectures running Windows Server 2003 are also discussed, emphasizing the support built into Windows Server 2003 Dataccenter. It also introduces tools like WSRM, available in Windows Server 2003 Enterprise Edition, that help in managing heterogeneous workloads on large scale NUMA machines.

Mark Friedman, Demand Technology

Mark Friedman is the founder and President of Demand Technology Software, headquartered in Naples, FL. The company develops tools for Windows performance monitoring and capacity planning. His previous experience includes senior technical and management roles at Datacore Software, Landmark Systems, Morino Associates, and StorageTek.

Mark has broad experience in IT corporate and business development. He edited and published a monthly newsletter, Storage Management, from 1993-1999. He founded OnDemand Software in 1994 and was Chairman of the Board of Directors of that company when it was sold to Seagate Technology in the spring of 1996. OnDemand developed and marketed network utility software, including the award-winning WinInstall software distribution package. In addition to his direct involvement in these start-up technology businesses, he has been a private investor in several software start-ups and served as a member of the Board of Directors for Terascape from 1996 to January 2000 when the company was acquired by EMC Corporation.

Mr. Friedman's experience in commercial Information Technology spans twenty-five years with Fortune 100 corporate data centers, government, hardware vendors and commercial software houses. He is a recognized expert in computer performance, disk and tape performance, storage management and distributed storage management. He was a regular contributor to Enterprise Systems Journal and is in demand to speak at Computer Measurement Group, SHARE, GUIDE, Storage Management User Groups, the RAID Advisory Board, and meetings of other professional organizations. Mr. Friedman's training seminars, lectures and published work are highly regarded for their technical quality and depth, and he is esteemed for his ability to communicate complex technical topics in plain, concise terms. He holds a Master's degree in Computer Science. His book, Windows 2000 Performance Guide, was published by O’Reilly Associates in February, 2002.

2:15 – 2:30

Break

2:30 – 3:30

Getting to Know WebLogic Server Clusters

Overview of configuration and benefits of using WebLogic Clusters in order to achieve Reliability, Availability, and Scalability

in application server architectures.

Ellen Johansen, BEA

QA Manager for BEA's WebLogic Server Division. 5 years experience of working on WLS Clusters in load and stress environments. 20+ Years industry experience.

3:30 – 4:30

Looping for Performance - A Tuning Methodology

In the computer performance arena we often think of “Looping” in a negative sense. But is all looping really bad? The loop control structure is actually the most often used control structure in programming today. This paper suggests a use of the loop structure in our daily performance analysis to find the largest users of a system resource. In our example we study the use of this technique as applied to identifying and reducing the CPU portion of your Total Cost of Ownership (TCO). It can also be applied to your I/O and memory resources with similar success.

Greg Scriba, BMC

28 years in the industry - Positions in Operations, Systems, Performance and Capacity Planning. Currently working for BMC Software with the MAINVIEW for Performance Assurance Product line, that solution was called Best/1 from BGS in prior life. I have worked in the financial, manufacturing, car rental and hardware vendor industries.