 |
 |
|
|
 |
|
|
||
|
|
|
|
|
|
|
||
 |
 |
 |
|
|||||
 |  |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
||||||||||||||||||||||||
|
Connecticut Computer Measurement
Group
Our June meeting marks the beginning of the summer drive. Summer is a time to get out, away from the office, and enjoy. What better way to enjoy life than to take part in a fabulously full day of terrific topics, engaging speakers, and educational opportunities at a fraction of what you or your company would typically spend for a workshop or seminar. Connecticut CMG has been very fortunate over the years to have industry honored speakers come from all over the country to this Northeast corner to share their knowledge and special expertise on the most pressing needs of our work. This meeting achieves our highest goals in bringing to our members both leaders in our field and topics that challenge all of us everyday.
8:00-8:30 Server Concepts Glenn Anderson – IBM Rob Breen – Aetna, Inc. 10:00-10:15 A Multi-Faceted Relationship Glenn Anderson – IBM Sid Soberman – H. W. Wilson 11:30-12:30 Michael Salsburg, Ph.D. – Unisys Corp. 1:45 - 2:00 Yiping Ding, Ph.D. – BMC Software, Inc. 3:15 – 3:30 -------------------------------------------------------------------------------- About the Speakers Glenn Anderson is currently a Consulting Instructor with the zSeries Curriculum of IBM IT Education Services. In addition to delivering classes on performance, e-business and web technology, Glenn is a popular speaker at large systems conferences throughout the world. Glenn is also constantly engaged in creating new conference formats and other educational vehicles in partnership with the IBM Server Group. During his 30 years with IBM, Glenn has held positions in sales, technical support and management. His current areas of research, focus and study include WebSphere for z/OS, e-business security, the z/OS Workload Manager, and all aspects of zSeries and Parallel Sysplex performance. Glenn holds a B.S. in Computer Science from the University of Missouri - Rolla. In his private life, Glenn is active in Chicago-area theater as a member of an improvisational acting troupe. He also performs original stand-up comedy, both in clubs and occasionally in the classroom. Rob Breen is a working musician with a day job in computer programming. He holds a Bachelor of Music degree from Boston University. In 1984, Rob began his career at the Aetna in Middletown, eventually moving into the technical support area where he polished his assembler language skills as an IDMS DBA and software engineer. At the mainframe level, Rob made the transition to DB2 support and since he was the only guy in the shop who could spell “grep” found himself supporting UDB platforms on both AIX and Solaris. An interest in solving performance problems led Rob to join the midrange systems team in the Performance and Capacity Planning area in 2001. Sid Soberman is currently Director of Capacity Planning at H. W. Wilson Company. He has over 32+ experience in the IT field. The last 22 years has been spent on application integration, performance management and capacity planning studies. His experience has ranged from mainframe to open system environments. Sid has been a member of CMG since 1984 and has assisted with the conference and program committee the last 10 years. Recently, he has been the subject area chair for Networking for the annual CMG International Conference 2004. Dr. Michael Salsburg is currently a director within the Chief Technology Office for Unisys Corporation’s Systems and Technology group. He is responsible for the strategic technical direction of the overall server product line, with a focus on System Performance and overall Architecture. He received his Doctor of Philosophy in Mathematics from Drexel University (Probability and Statistics) in 1992. Dr. Yiping Ding is an Architect/Scientist with BMC Software in Waltham, MA. He has been with BGS/BMC for more than 12 years and is responsible for BMC's performance data analysis and prediction product development. His technical interests include performance modeling and statistical data analysis techniques. He received his Ph.D. in Computer Science from the University of Connecticut in 1991. -------------------------------------------------------------------------------- Agenda Abstracts Glenn Anderson – IBM IT Education Services Are you confused when hearing about work managers, enclave servers, queue managers, server managers, or routing managers? Glen will explain the concepts of the different server or manager types, discuss their differences, and show you the areas where these concepts are implemented to be beneficial in terms of performance, efficiency, and manageability for your system or sysplex. Examples will include WebSphere, CICS and DB2. 10:15-11:30 WLM and z/OS WebSphere Ver 5: A Multi-Faceted Relationship Glenn Anderson – IBM IT Education Services The WebSphere for z/OS Version 5 runtime environment interacts with WLM in three key areas. First, WLM routing services are used to balance work across a parallel sysplex, and within the WebSphere server instances in a WebSphere node. Second, WLM application environments are then employed to manage the server regions where the EJBs and Web Applications run. Then, finally, WLM classifies the WebSphere work into service classes using a variety of available qualifiers and the work runs using WLM Enclave Services. This fast-paced session will explore this multi-faceted relationship between WLM and WebSphere for z/OS, focusing on what the WLM performance guru needs to know to maximize these interactions. Open Systems Track Rob Breen – Aetna Have you ever found yourself in a tight spot where you needed a quick and accurate UNIX performance measurement that the tools at hand simply couldn’t provide? My shop has been there a time or two, and over the years we’ve developed a suite of very useful and reliable tools to meet those emerging needs. Using commonly available scripting languages and presentation tools, we’ve been able to sift critical bits of information from netstat and vmstat commands which have allowed us to detect system problems as they were developing and reduce problem resolution times by focusing quickly on problem areas. Let me share with you some of the practices and philosophy we’ve followed through the evolution of our library of homegrown monitoring tools for Solaris and AIX. 10:15-11:30 Switching Hardware Platforms: From Solaris to Linux and Surviving Sid Soberman – H. W. Wilson Much has been written concerning hardware migration from expensive RISC servers to commodity Intel servers. This presentation describes one company’s hardware platform migration from SUN to Intel because of the use of Linux. This critical decision occurred one month before production deployment of a new web subscription search service. A load-testing tool was used to measure significant application, database design changes and hardware platforms during development and beta stages. Sid describes the process and talks to the testing methodologies used as well as the surprising results. Combined Sessions 12:30-1:45 High Performance Computing: Workload Characterization and Modeling Michael Salsburg, Ph.D. – Unisys Corp. High Performance Computing (HPC) workloads impose very specific demands on computer systems. In some ways, they have guided the evolution of computer architectures into new areas of research, development and utility. There has been and continues to be significant research into key characterizations of the memory/CPU interplay. This paper explores workload characterization and modeling issues that go beyond this detailed level of analysis. A high level modeling approach is needed to evaluate efficiencies in highly parallel systems. For applications that demand unusually high storage data transfer requirements, we need to understand the fundamental characteristics of the application's I/O workload. Many of the more robust applications require significant communication fabrics and can realize higher efficiencies with reduced latencies within the fabric. This paper presents HPC workload characterization and modeling concepts to explore issues such as these. 2:00 – 3:15 Business Metrics and Capacity Planning Yiping Ding, Ph.D. – BMC Software One of the major goals of capacity planning is to ensure that the service level objective of business activities is met. Identifying the relationship between business metrics of interest (BMIs) and system performance metrics is the first step towards the goal. This presentation will define the characteristics of a good BMI and give techniques for identifying it from your business user’s processes. Yiping will discuss analytical tools used to decide if your business metric matches the system resource consumptions and examples of how increasingly fuzzy BMIs can make great capacity plans. |
|
||||||||||||||||||||||||||
 |
|
