Optimizing z/OS system and application performance today is all about having data as close to the processor as possible when it’s needed. Despite some incredibly fast I/O times today, memory is still orders of magnitude faster and (still) the only good I/O is no I/O. Keeping data in memory can not only improve application performance but reduce CPU consumption as well. Fortunately, many systems today have sufficient memory to make potentially significant improvements. But memory is not infinite so how do you find the best opportunities for improvement? In this session Scott Chapman will discuss data sources and methods to find, evaluate, and implement I/O reduction opportunities. You’re sure to leave with good ideas that will help you eliminate unnecessary I/O and improve application and system performance.

There are many reasons to "open up" a WLM service definition. Some changes need to be made for tuning, others for clean-up, and still others because the goals are outdated. Examination needs to be performed for planned environmental changes or changes to business priorities. The point is that there are many reasons to open up the WLM service definition for examination and potential changes. During this presentation, Peter Enrico reviews the different reasons to roll up your sleeves and analyze your WLM service definition. During this presentation, you are sure to learn a lot about a wide variety of areas related to WLM management on z/OS.

Inside your mainframe there are physical CPUs, but in almost all cases z/OS gets logical CPUs to run work on. This distinction is sometimes confusing to people, especially in relation to how it can affect performance and available capacity. In this session we’ll discuss logical CPUs, physical CPUs, and the importance of understanding LPAR configuration options that can impact performance. After attending this session you should have a clear mental model of logical versus physical CPUs as well some ideas about how to optimally configure your important LPARs.

During this session, Peter Enrico and Scott Chapman of Enterprise Performance Strategies, Inc. (EPS) will discuss a variety of z/OS performance measurement, analysis, and tuning techniques that may not be commonly known or are not often discussed. The key objectives of this presentation will be to provide the attendees with information they can bring back to their shop and conduct some analysis or tuning exercises. A secondary objective of this session is to help the attendees learn more about the z/OS environment, and how things work. This session is sure to be highly educational!

This webinar will walk through and explain several reports that will be useful when evaluating the CPU capacity and usage of coupling facility processors. Dynamic dispatch, thin interrupts, and the differences of physical CF processor utilization compared to virtual processor utilization will be discussed by Peter Enrico.

Over the last several z/OS versions and machine generations, things have changed in terms of memory sizes and z/OS management of all that memory. During this webinar, Scott Chapman will discuss z/OS large page support (1MB and 2GB) and why using them is still a best practice. Also, details about how z/OS has changed to better manage large memory will be discussed. Things aren’t the same as they were 10 years ago!

In this webinar Scott Chapman will walk through and explain several reports that will be useful when evaluating the usage of parallel access volumes (PAVs). The concepts and reasons for PAVs will be discussed; then, some key reports will be reviewed to help analyze the effectiveness of the usage of the PAVs in your z/OS I/O subsystem.

This webinar will walk through how XCF management has been greatly simplified with automatic transaction class management. Significant changes were made to z/OS 2.4 so that transport classes are no longer needed to segregate messages by size or application.

During this webinar Peter Enrico will discuss all these changes and what needs to be done to take advantage of these XCF enhancements.

During this session, Peter Enrico and Scott Chapman of Enterprise Performance Strategies, Inc. (EPS) will discuss a variety of z/OS performance measurement, analysis, and tuning techniques that may not be commonly known or are not often discussed. The key objectives of this presentation will be to provide the attendee with information they can bring back to their shop and conduct some analysis or tuning exercises. A secondary objective of this session is to help the attendee learn more about the z/OS environment, and how things work. This session is sure to be highly educational!

There are many reasons to "open up" a WLM service definition. Some changes need to be made for Tuning, others for clean-up, and still others because the goals are outdated. Examination needs to be performed for planned environmental changes or changes to business priorities. The point is that there are many reasons to open up the WLM service definition for examination and potential changes. During this presentation, Peter Enrico reviews the different reasons to roll up your sleeves and analyze your WLM service definition. During this presentation, you are sure to learn a lot about a wide variety of areas related to WLM management on z/OS.

This presentation will provide a comprehensive overview of parallel Sysplex in a z/OS environment. The attendee will learn the basic concepts of parallel Sysplex and data sharing. Covered in the presentation will be an introduction to coupling facility and its resources, coupling facility structures and how they are used, and exploiters of the coupling facility. Also covered in the presentation is how data sharing actually works. While billed as a rookie session, this presentation will even teach the seasoned z/OS professional a few new things.

Do you remember when “large” memory was measured in single-digit GBs, or maybe even MBs? Yeah, I feel old too! It’s a different world today! With z/OS 2.5 now supporting up to 16 TB of real storage z/OS memory management has had to evolve to keep pace. This support for large memory sizes can transform application performance but too few sites are taking advantage of it. Come to this session to learn about how memory management has changed in z/OS, why large pages are important, and get some ideas for how you can leverage large memory to improve the performance of your systems.

Pivotor is a premiere z/OS performance reporter for intelligent reporting of your SMF data, as well as other sources of measurements. There are a wide variety of features and functionalities that are different from other comparable products. In this webinar Scott Chapman will highlight many of these features, and you are sure to see why Pivotor is a premiere product for z/OS performance reporting.

Performance management for z/OS systems is a well-established field, and z/OS has a wealth of performance data and tools to help with that. However, sometimes organizations make decisions or have practices that limit their ability to effectively manage z/OS performance.

In this session Scott Chapman will explore some of those anti-patterns, from the perspective of the easy things you can and should do to make it easier for you to understand and manage z/OS performance.

During this webinar, WLM expert Peter Enrico will discuss some of his key analysis exercises to determine if your WLM is set up optimally. Along with these analysis exercises, Peter Enrico will provide a number of recommendations that are sure to benefit the performance and resource optimization of most z/OS environments. After you attend this webinar, be prepared to roll up your sleeves and get to work because this webinar will be full of useful and relevant information.

How do HiperDispatch High, Medium, and Low pool processors influence the performance in a z/OS environment? What does it mean when a logical processor is designated as a high pool processor, and how does a high pool processor impact the performance of a z/OS LPAR differently than a medium pool processor? In this webinar Scott Chapman will discuss HiperDispatch and how various HiperDispatch pooling influences the performance.

This presentation discusses the concept and usage of calculated intensity measurements. When there are many of a particular resource or workload to be evaluated, intensity is a calculated measurement to help highlight which of the many are impacting performance the most. For example, say you have a DASD farm with 3000 logical volumes - certain intensity measurements can help prioritize which of these many logical volumes are impacting performance the most. 

Managing DDF workloads can be somewhat tricky because DDF can be quite variable. You may have an application where each row inserted is an extremely short transaction. Or you may have QMF users who submit queries that run for hours or even days. And you probably have a lot of DDF work that falls in-between but which varies in both importance and intensity. Historically, the full range of WLM goal types and controls were available to help manage this challenging workload. But now, in some situations, that’s no longer the case. If you use DDF and want to better understand the WLM options for managing the work, and especially if you’re using (or may use) high performance DBATs, you should attend this presentation delivered by Scott Chapman to better understand how to take control of DDF.

In this webinar Scott Chapman will provide a brief overview of System Recovery Boost (SRB) and then discuss the broader implications on performance and performance reporting. As a reminder, during a speed boost, sub-capacity CPs will run faster, and a zIIP boost will provide additional capacity by allowing CP work to run on zIIP processors. SRB can improve the return-to-service time for various planned or unplanned “recovery” activities, but there are some caveats to be aware of, especially if you’re doing performance analysis. Turns out some of the data will lie to us during a boost period.

This presentation will walk through, and explain, several reports that will be useful when evaluating the primary processor cache measurements on the z16 processor. There will be a review of the basic concepts and usage of the processor caches and then it will show which reports and measurements should be used to assess the effects of processor caches in the z16 environment.

In this webinar, Peter Enrico will provide some deeper insights into the virtualization of the processor caches on the z16 processors.

The Mainframe Service Provider (MSP) industry has been growing at a steady rate with many companies engaging with MSPs to outsource the management of their mainframe environments. Companies do this for a wide variety of reasons including staffing flexibility, environmental and maintenance costs, and growth flexibility. With an interest in capacity planning and performance management, Scott Chapman and Peter Enrico of Enterprise Performance Strategies have worked with both MSPs and clients who outsourced their environments to MSPs. They have also worked with clients that were just mulling over the idea of doing so. Their experience has enabled them to develop a unique level of insight.

During this presentation, Scott Chapman and Peter Enrico will discuss common pitfalls and issues related to the performance management of a mainframe environment that has been outsourced. If your mainframe environment is currently outsourced, or you are thinking of outsourcing, and want to learn more about these pitfalls, then this webinar is for you.

In the first part of this webinar series, Scott Chapman and Peter Enrico discussed some performance management pitfalls outsourced customers need to be aware of to ensure they are being provided for optimally by their Mainframe Service Providers (MSPs).

During this presentation, Scott Chapman and Peter Enrico will discuss some performance management reporting that should be monitored to help ensure the services they are being provided are to the greatest benefit to the client customer, and not just to the MSPs.

The WLM component of the z/OS operating system 'wakes up' on a regular basis to adjust system settings to meet goals. What is happening when it 'wakes up'? During this presentation, Peter Enrico will explain how WLM works and the flow of the WLM algorithm decision making process. This is an interesting presentation, and the attendee is sure to learn a great deal about WLM.

The SMF 98 and 99 records can provide performance data at a much finer time granularity than we normally get from the RMF and SMF data: down to 2 second intervals for some data!

In this webinar, Peter Enrico will show how we can leverage these under-appreciated records to gain deeper insights into performance issues that may be relatively short-lived.

Many customers only replace their mainframe every 3-5 years, so properly planning for what machine to go to next is not something they have a lot of practice at. Many simply rely on their vendor to make a good suggestion. But the wise customer will consider multiple factors before making a final decision.

In this session, Scott Chapman will share the insight he has garnered from years of planning processor upgrades. He’ll explain the basics of running zPCR, why you want to do so, and how you can use the output. Preparing for processor speed changes will be discussed. Finally, some thoughts will be provided about understanding the performance changes after the upgrade.

During this session, Peter Enrico and Scott Chapman will discuss a variety of z/OS performance measurement, analysis, and tuning techniques that may not be commonly known or are not often discussed. The key objective of this presentation is to provide the attendee with information they can bring back to their shop and conduct some analysis or tuning exercises with. A secondary objective of this session is to help the attendee learn more about the z/OS environment and how things work. This session is sure to be highly educational!

The SMF 14 and 15 records are some of the very first SMF records that IBM introduced way back during the dawn of SMF records recording on the MVS platform. Many installations still process and examine these records. Some installations process these records because their data set I/O activity reporting programs are ancient, while others use these measurements because in these records, we find certain measurements not found in any other SMF records.
In this webinar, Peter Enrico will explore these SMF record types, their history, their measurements, and why you may or may not want to use these records to better understand the file I/O activity of your z/OS environments.

Typically, the Coupling Facility lock structures are the most expensive structures in your data sharing environment. In this context, expensive means that lock structure operations typically consume more MSUs on your z/OS images than other types of structures.

During this presentation, Peter Enrico will discuss coupling facility lock structures and lock structure performance, so you have the insights you need to help improve performance and reduce MSUs.

As always, z/OS processor measurements are critical during any z/OS performance analysis and tuning exercise. During this webinar, Scott Chapman will explore the WLM workload processor measurements in the SMF 72 record. Specifically, we will look at some of the key processor measurements available in the SMF 72, subtype 3 record and then explore calculating and using processor measurements at the service unit level, at the CPU seconds level, at the application percentage (APPL%) level, and finally at the workload utilization level.

During this presentation, Peter Enrico will go into more SMF 30 address space CPU time measurements. In particular, Peter will explain more about the concepts of various types of dispatchable units and the concepts of independent and dependent enclaves. Peter will then relate these concepts back to the address space measurements available in the SMF 30 records.
This webinar is scheduled for 30 minutes, but after the webinar, Peter Enrico and Scott Chapman will be available for a post-webinar Q and A.

Address space activity measurements are recorded in the SMF 30 record. There are so many measurements in the SMF 30 records that we could probably do an entire webinar series on just the SMF 30 record.

During this webinar, Peter Enrico will both introduce the SMF 30 record and then explore the key processor measurements you may be interested in. Also discussed will be some ways you may want to examine and use these address space CPU measurements.

Serialization of resources on z/OS is such a fundamental need, that locking is one of the foundational concepts of the MVS platform. There are CMS locks, local locks, CML locks, GRS latches, suspend locks, spin locks, and more.

During this webinar, Peter Enrico, Scott Chapman, and Bob Rogers will explore the different lock types and any measurements available to help us understand the locking activity of our z/OS environments.

Of all the z/OS SMF records available to us, the SMF 70 record is probably one of the most critical records to understand and use. Within the SMF 70 record, the processor dispatch interval measurements are the most important.

During this webinar, Peter Enrico will explore the SMF 70 processor dispatch measurements, and the formulas that these measurements are used to calculate both physical and logical processor utilizations. Oh... and what is the conceptual difference and usage between physical and logical utilizations? This difference will also be discussed.

This presentation will provide a comprehensive overview of the z/OS Workload Manager (WLM). The attendee will learn the basic constructs of WLM, how WLM manages workloads and optimizes resources in the z/OS environment.

The attendee will walk out of this presentation with not just a working knowledge of WLM, but also a few hints, tips, and exercises they can use back at their installation.

On the z platform there are more processor measurements than any other computing platform. What are all these processor measurements actually measuring, where do they come from, and how can they be used?

During this presentation Scott Chapman will introduce to you the many processor measurements available, and show you how many of them can be used during any processor performance analysis or capacity planning exercise.

Today, not all, but a great many z/OS environments have plenty of processor memory (i.e., storage). Whether your systems are storage rich, or teetering on the edge of paging, it is always important to occasionally look in on your processor storage usage.

In this webinar, Scott Chapman will explore some key processor storage measurements to help you understand what types of storage frames your systems have configured and are using. Also discussed will be some key measurements you should regularly monitor, regardless of if you are z/OS processor storage rich or poor.

The SMF 98 and 99 records can provide performance data at a much finer time granularity than we normally get from the RMF and SMF data: down to 2 second intervals for some data! In this webinar, Scott Chapman will show how we can leverage these under-appreciated records to gain deeper insights into performance issues that may be relatively short-lived.

The concepts and formulas for External Throughput Rates (ETRs) and Internal Throughput Rates (ITRs) are the foundations for IBM's evaluation of the capacity of the mainframe processors and of the Large System Performance Reference (LSPRs) tables. During this session, Peter Enrico will review the concepts of ETRs and ITRs, and he will show you how they are used to evaluate not just processor capacities, but also how they can be used to evaluate changes to your workloads. This session will also provide insights into IBM's LSPRs which are used to represent IBM's assessment of relative processor.

WLM response time goals: we all love them if only because velocity goals are difficult to understand and maintain. But it turns out that response time goals have their own issues as well. In this presentation, we’ll discuss response time goals and when you do and don’t want to use them. We’ll also compare average and percentile response time goals and when you might want to use each. Spoiler Alert: average response time goals can be useful in the modern mainframe environment!

SMF data is a key input to mainframe performance analysis. Yet, we often see installations that are not recording useful records or recording records in a way that makes performance analysis more difficult than it needs to be. In this webinar, we will discuss some of these anti-patterns and how a few simple SMFPRMxx changes can be useful.

CPU overhead caused by the programming anti pattern of updating data that’s too close to the instruction stream has long been a known problem. IBM came out with a nice formula that highlights time intervals where that practice may be causing significant overhead. But that’s only part of the battle: the next question is how do you find the possible culprits so you can remediate them?

At EPS we have years of experience helping people understand the performance of their z/OS systems, in part by visualizing the performance details that are stored in the SMF data. You probably won’t be surprised to find that we’re quite interested in data visualization as a discipline.

In this webinar, Scott Chapman and Jamie Novotny will explore the psychology behind data visualization, discuss some visualization best practices, ridicule examples of poor visualizations, and explain how Pivotor reports work to avoid obfuscation and promote understanding.

Before doing any sort of performance valuation, it is super important to first understand both your managed resources and your workloads. During the webinar, Peter Enrico will go through a list of important details you should learn about both your resources and workloads to help you prepare for any performance analysis assignment.

For z/OS 2.5, the WLM Service Definitions Coefficients (SDCs) have finally been eliminated. For years, we have been recommending CPU and SRB set to 1, and MSO set to 0. As of z/OS 2.5, the hardcoded values will be CPU and SRB to 1, MSO and IOC to 0. For some installations, this will require tweaking of the durations of their multiple period service classes. During this webinar, Peter Enrico will discuss this change in more detail. He will also provide an simple exercise so that will help you prepare for z/OS 2.5 by adjusting your period durations now. So, join this webinar knowing your current service definition coefficients and period durations.

SRB is one of the more interesting things that IBM has introduced with the z15 and certainly can help certain installations, shut down, start up, and recover faster. But what are the practical implications of using SRB? How does enabling the different flavors of SRB influence both the performance and measurement of the systems and even potentially the systems that aren’t being boosted?

Join Scott Chapman in this session as he explores SRB and discusses why System Recovery Boost is like Solid Rocket Boosters: exciting and getting you where you’re going quickly but needs to be handled with some care.

Most people use Pivotor on a day-to-day basis to understand and manage the performance of their z/OS systems. But sometimes a longer-term perspective is needed to answer questions like…

• "When did this start happening?"
• "How much worse was yesterday than last month?"
• "When would be a good time to schedule this work?"
• "Is our long-term CPU consumption trending upwards?"

Ah... the SYSTEM and SYSSTC service class. These are two foundational service classes that all installations have defined, regularly use, and are in need of evaluation. During this webinar, Peter Enrico will give a primer of the SYSTEM and SYSSTC service classes and provide some general performance guidelines.

The SMF 113 record has been around for while now. It is expected that a new processor will be announced by IBM sometime in, the somewhat, near future. Understanding, evaluating, and using the SMF 113 processor cache measurements is always critical when evaluating your current processor or considering an upgrade.

During this presentation, Peter Enrico will provide a back-to-basics presentation with the objective of reminding everyone why the SMF 113 processor cache measurements are interesting for analysis.

The ability to have and management workloads in multiple period service classes has been around since the very first days of WLM. In fact, multiple period service classes are a continuation of the ancient multiple period performance groups of early MVS. However, the using and managing multiple period service classes has evolved. During this webinar, Peter Enrico will review and discuss some advantages of using multiple period service classes in the year 2021.

Several years ago, IBM added an option to record instruction counts in the SMF 30 data. They almost immediately realized that the instruction counts were skewed by I/O interrupts and so the recommendation was to not use them. More recently, IBM has started to say that maybe you can derive useful information from the instruction counts if you just discard the bad measurements. Does this make sense? Are there insights to be gleaned from those instruction counts? Can you use them to find inefficiencies? Or does the natural variability introduce too much noise to make sense of any signal that may be lurking in the data? In this session, we will explore what we have discovered by looking through data from multiple customers that have enabled the instruction counters in the SMF 30 data.

Are you new to the mainframe and need to understand the performance of your z/OS systems? Have you seen some basic z/OS performance measurements but don’t understand the source of those measurements? Or maybe you’ve seen some measurements that don’t sound like measurements that you’re familiar with on other platforms?

If you’ve answered “yes” to any of those questions, check out this presentation, given at STLCMG by Scott Chapman to lead you through some of the more important z/OS performance metrics and explain where they come from and why they’re important.

There is a plethora of z/OS performance metrics available. Some are easily understood and their applicability is well-known. Others are not so obvious. The more difficult question can be “what is a good value for this metric?” For some metrics, there are no “right” or “wrong” values as they are workload-dependent, but it is still interesting to understand the range of values commonly seen. Situations that generate metrics outside the normal range might be worth investigating.

In this presentation given at STLCMG by Scott Chapman, he gives a tour through some interesting z/OS performance metrics, both familiar and not-so-familiar. He will provide a sense for what values other customers are seeing for these metrics as well as some modern rules of thumb where applicable.

Not all performance topics and recommendations are simply cut and dry. Many are controversial. These are the recommendations that tend to generate discussion among peers, need careful consideration, or may depend on which ‘performance camp’ you belong to. During this presentation, Peter Enrico and Scott Chapman will explore some of these recommendations.

In this presentation given at STLCMG by Scott Chapman and Peter Enrico, you are sure to learn something new. The goal of this presentation is to provide you a deeper understanding of these performance topics that don't always have a simple answer.

Are you new to performance analysis? Or maybe your manager has tasked you with a performance analysis assignment. Or maybe you just want to make sure you are prepared for any performance problem that comes you way.

This presentation might not discuss how to do one specific performance analysis. Instead, during this presentation, Peter Enrico will discuss some very basic tasks you need to do to prepare for any performance analysis that may come your way.

Service classes can have multiple periods. Why is this? Where does it make sense to use multiple periods? How should you set the period duration? In this webinar, you will learn when it makes sense to use multiple periods for a service class and how to set intelligent durations for the periods. Intelligent use of multiple period service classes can help overall system performance and throughput.

WLM was introduced in the early 1990s and most sites most sites implemented goal mode in the late 90s or early 2000s. A lot has changed since then. New WLM features have been introduced. Options and features are enabled (or not) either when defining a new policy or when a new feature becomes available but then never reviewed for applicability in the face of significant changes in the hardware and software environment. Just like regularly reviewing your goals, revisiting your WLM options periodically is a good thing as well. Join Scott Chapman in this session has he reviews the WLM options and shows how one long-standing recommendation is now perhaps skewing results in a counter-productive fashion.

Mainframe customers have had a choice of processor speeds for years. This flexibility has been somewhat limited by the number of sub-capacity engines that can be configured. But now with the z13, up to 30 sub-capacity engines can be used in a single machine, which opens the consideration of sub-capacity engines to more sites. What are the advantages to using sub-capacity engines? Does your LPAR configuration matter? What metrics should be examined to determine if more/slower CPUs is a good fit for your workload? Is it possible that this choice might impact your software bill, either today or in the future?

Come to this session with Scott Chapman to learn why more/slower CPUs may be a better fit for many environments and how to determine what workloads might be at risk for moving to slower CPUs. Representative configurations will be examined and discussed.

Measuring CPU consumption is a critical piece of both performance analysis and capacity planning. There are lots of different ways that CPU consumption can be expressed such as CPU Seconds, MIPS, MSUs, Service Units and still more. If you are confused as to what these terms mean and how they relate to each other, then this session is for you.

Join Scott Chapman on a walk through the possible CPU measurements, conversion between those measurements, and when you might use each.

Should you enable Simultaneous Multithreading (SMT) on your z13, z13s, or z14 for your zIIP workloads? That is an interesting question that doesn’t have an easy answer. If you do enable SMT, you should be prepared to review a number of SMT-specific measurements, as well as your standard application measurements. Unfortunately, these are some of the most confusing, misunderstood, and poorly documented measurements in all of the SMF data.

In this presentation, Scott Chapman explores the meaning and derivations of the SMT measurements and then presents existing measurements that may be useful in determining if SMT should even be explored. A decision tree for deciding whether to stay in single-threaded mode or try SMT-2 will be introduced.

This is one of Scott's "What I Learned This Month" columns that he wrote for CMG while at AEP.

In this WILTM, Scott compares the CPU consumption of Java 7 vs. Java 6 for some particular workloads, examines the overhead that 64-bit adds and how additional JVM options mitigate that impact. While nobody today should be running Java 6 or Java7, we're including it here for historical reference and because it's an interesting example of how parameter changes can influence performance He also utilizes the SMF 113 measurements to validate and explain the changes in CPU consumption.

This is one of Scott's "What I Learned This Month" columns that he wrote for CMG while at AEP.

In this WILTM, Scott tests out the Flash Express feature on a new zEC12 machine. While current iterations of "Virtual Flash Express" is much different than the hardware that was introduced on the zEC12, we're including this as a historical reference. Today's Virtual Flash Express will of course be much quicker.

This is one of Scott's "What I Learned This Month" columns that he wrote for CMG while at AEP.

In this WILTM, Scott discusses the philosophical differences between platforms, specifically around defining database columns as either CHAR or VARCHAR. An issue with DB2 VARCHAR columns used in indexes is explored. While this is a relatively old at this point, we're including it here in the hopes that it may help somebody else.

This is one of Scott's "What I Learned This Month" columns that he wrote for CMG while at AEP.

In this WILTM, the benefits of DB2 compression are measured to see if it still makes sense in today's world. Note that "today" was 2012. We're including it here because we think it might still be of interest to people, although the result might be different if he did the investigation today.

This is one of Scott's "What I Learned This Month" columns that he wrote for CMG while at AEP.

In this WILTM, a performance problem is solved by enabling the PDSE "buffer beyond close" option. We're including it here because we think it might still be of interest to people, after all the only good I/O is still no I/O!