In contrast to what you may be used to on a UNIX system, systems management on the mainframe goes far beyond just managing a few users. Systems management covers an extensive range of policies, procedures, and tools. We will divide systems management into essentially arbitrary disciplines that, nonetheless, reflect the division of responsibilities in many large IT installations. Furthermore, the disciplines we are going to use also map to the areas in which many tool providers specialize.
We will distinguish these disciplines:
The following subsections provide an overview of these disciplines. For each discipline, there is a reference to a chapter with more detailed information.
12.2.1 Availability management
Availability management covers all the strategies and activities that have the prime objective of ensuring that systems and applications are available to the target end users for their intended use. The more central your systems are to your ability to do any business at all, the greater are the demands on availability.
Running applications on zSeries gives you hardware for which the mean time between failures (MTBF) of a machine is currently measured in decades, and Linux is itself widely valued as a solid operating system. Linux on the mainframe has the potential for excellent availability and we argue that it can be managed to leverage that potential.
With that foundation, the risks involved in quickly deploying a new application are centered on what you have most control of: the application itself.
Availability is always a high priority, but it is not the only issue in systems management. There are other aspects, such as security and performance, that compete for importance which might require trade-offs with availability. How to rank these aspects and where to strike the inevitable compromises for a particular installation can be inferred from the business objectives that the installation has been set up to serve.
While StoreCompany also benefits from good availability of a new Internet sales project, its prime concern is to ensure that the new project does not compromise existing procedures and the integrity of its database. Security issues and isolation requirements for the new project thus override the new project's availability considerations.
Chapter 13, "Availability Management," addresses the question of which level of availability is appropriate for a given project or application. It also points to some common availability management procedures that have unique considerations in an environment with Linux on the mainframe.
12.2.2 Data management
Data management covers all efforts to have the data available that are to be processed by systems and applications. In a Linux-on-the-mainframe environment, the z/VM virtualization technology can help you to share data between Linux images. In particular, both an image and a hot standby that are ready for failover can share access to data.
A company's business data are a major and vital asset that often cannot readily be recreated. Consequently, many firms have substantial procedures in place to ensure that backups are kept in safe places in case of damage to the currently active set of data. Many governments also have laid down rules on privacy and on record-keeping that have to be covered by the data management procedures.
Our focus is on the logical aspects of managing the data, rather than the physical aspects. We discuss the physical devices only to show what is there for you to use and what consequences the choice of physical devices might have on your options for handling the data. Some of what we discuss under data management is also classified as storage management.
Data management activities typically include planning for space requirements for increasing amounts of data. It also involves aspects of security management in ensuring that data are visible and accessible only to programs and persons with specific authorization to the data.
Data management needs to address the interrelationships of sets of data. Sometimes it is critical to have sets of files backed up with the content they had at the same point in time. Such backups require significant coordination with application availability.
Data management is tightly coupled to availability management. If the processor is up and the application is ready to do work, yet the data to be processed are inaccessible, then the end user sees a broken system. The mainframe allows you to attain high data availability, for example, through redundant data access paths and efficient backup tools.
Chapter 14, "Data Management," explores this discipline with respect to Linux on the mainframe.
12.2.3 Security management
The challenge of security management is to make a system as secure as possible without unduly affecting the legitimate user.
Security management protects systems from intended or accidental damage and thus supports availability. Security, however, often also conflicts with availability because it restricts accessibility and deploys barriers that only intended users and applications can overcome. Forgotten passwords can mean that StoreCompany customers temporarily cannot access their accounts with the StoreCompany Internet sales application. Trade-offs have to be made, depending on the overriding concern.
Security issues must be considered when conceiving the layout of a system, so we covered much of the security management discussion in Part 3, "Is Linux on the Mainframe for Me?," which includes Chapter 8, "Security Considerations," and Chapter 9, "Setting Up Linux on the Mainframe." Because that discussion also includes operational issues, there is no separate chapter on security in this part.
12.2.4 Performance and capacity planning
For the sake of discussions in this book, those aspects of "performance" that are used to determine system health are covered under availability management. The performance management and capacity planning chapter addresses those activities that are less directly associated with running systems.
Performance management aims to understand the behavior of the IT infrastructure (such as hardware, software, networks, etc.) to assure that resources are used in an effective way and SLAs (Service Level Agreements) are complied with. Where the results of performance analysis call for changes to the system structure, the running and the availability of the system are affected. Here, performance management intersects with availability management.
Capacity planning takes the analysis of performance data one step further. The objective in capacity planning is to anticipate the impact of changes to workload and resources in the system. The first three disciplines focus on activities that work together to keep everything running smoothly under stable conditions. Capacity planning is about predicting the behavior of the system if something were to change. It is used to answer challenging questions such as, "What happens to response time if we put this new product marketing campaign up on System 3?"
Chapter 15, "Performance and Capacity Planning," explores this discipline with respect to Linux on the mainframe.
12.2.5 System administrator tasks
Rather than introducing a multitude of additional disciplines, we have grouped all activities that do not fall into the above four disciplines into a single fifth category. It includes various activities typically focused on the underlying software infrastructure and the hardware, performed by the most skilled members of the IT staff.
Time-to-market for new ideas is one good reason for having Linux on the mainframe. How this potential is leveraged depends on your key staff of administrators and the tasks that they perform.
Chapter 16, "System Administrator Tasks," delves into some of the cultural (organizational) challenges that occur when marrying the tightly controlled mainframe environment with the Linux mentality of exploiting the latest and greatest. The issue of change management is the common thread across Part 4, "Making the Most of Linux on the Mainframe," in general and Chapter 16 in particular. With Linux on the mainframe, we are now dealing with a large number of Linux images sharing resources where change and control are both key to success.
12.2.6 Putting it all together
The ultimate goal of systems management is to have an IT infrastructure available to support a business. Figure 12-2 illustrates how we see the different disciplines help to support this goal.
Figure 12-2. Systems management disciplines
Data management and security management directly support availability by providing meaningful content for processing and by protecting the system from damage. These three disciplines are basically operational and require constant attention.
Performance and capacity planning and the set of tasks that have been summarized under system administrator tasks provide the foundation to assure that the system will also be available in the future. These tasks can typically be done at scheduled times and do not require a person to be on call continuously.