Linux increases the complexity of mainframe Performance Management

Article Title: Linux adds complexity to mainframe performance management. Linux is a technology channel of the IT lab in China. Includes basic categories such as desktop applications, Linux system management, kernel research, embedded systems, and open source.

What is mainframe performance management? In the past, there was no difficulty in answering questions from mainframe manufacturers and non-IBM users: mainframe performance management is part of application and system management, focuses on device performance management rather than RAS (reliability, availability, and durability ). However, it will be hard for them to explain to CEOs or CFO why they are worried about performance management. Generally speaking, performance management, especially mainframe performance management, is always an afterthought action and lags behind in ensuring system startup and operation.

Note: The concept of performance management is vague for some users. Especially in the virtualization era, performance management is not about the performance optimization of each application, but to ensure that even the worst performance of all applications can meet the server-level requirements, that is, you can provide any end user with a response time of about 2/5 at the maximum load.

　　What are the changes in performance management today?

First, there is a new concept in performance management. In the business intelligence field, performance management allows CEOs to monitor and improve business processes, which is completely different from the old one. Therefore, when talking to CFO today, he may be totally confused about the performance management type you are talking about, even if he is very familiar with the importance of performance management.

In the past four years, mainframe costs have increased again due to the impact of mainframe Unix/Linux consolidation recommendations, resulting in lower total cost of ownership (TCO) for each application. However, with too much focus on TCO, mainframe server consolidation has changed its inherent performance. That is to say, the better the performance of your application, the more virtual machines run on the mainframe, and the lower the TCO of each application. Conversely, although user requirements or the number of applications are growing, the demand for adding capacity or new systems is decreasing. For many users, the stability of the mainframe is established. Only effective performance management can bring some benefits to the enterprise.
　
Therefore, the performance management of the mainframe becomes very important. At the same time, the influx of Linux applications (followed by the increase in Web access for end users) has fundamentally changed the performance management tasks of the mainframe. In the past, mainframe performance management focused on the internal performance of machines and seldom considered communications between virtual machines. Now, through Enterprise Application Integration (EAI) or Web and enterprise network communication outside the Organization, Unix/Linux applications can often access each other, and performance management must be "end-to-end. Network Communication and customer processes outside the mainframe and data center must be optimized, instead of focusing only on the communication status between virtual machines.

There is another area of change: software development. In the past, the pressure and data testing in the R & D process always appeared to be somewhat lagging behind, usually compressed, or simply not performed. In the same logic, few organizations automatically report problems detected during software operation to software testers or R & D engineers. However, the newly integrated Linux applications on the mainframe run very stably, and the independence between virtual machines is also good. This allows some organizations to apply the testing environment on virtual machines to actual machines, use it as the core application. This allows relevant personnel to easily report problems during software operation to testers. The improvement of software performance can save a lot of cost. In addition, the software can be returned to users for further operation through adjustment by testers, which speeds up the software performance adjustment and bottleneck problem solving.

In the lower half of this article, we will continue to reflect on the impact of these changes on mainframe performance management and how enterprise managers should respond to these changes.

[1] [2] Next page