WebSphere Commerce production Environment performance management
The E-commerce website with WebSphere Commerce as the core is a complex system composed of a variety of hardware and software products. In order to ensure the healthy operation of the system, the operations team needs to perform a comprehensive performance monitoring of the entire site and take the necessary proactive measures to maintain system performance. This article describes some of the best practices for performance monitoring and performance maintenance for WebSphere Commerce e-commerce Web sites.
Performance monitoring for WebSphere Commerce ecommerce web site
Performance monitoring is a variety of monitoring tools to understand the current application of the operation of the situation, to identify existing performance problems or may lead to performance problems potential risk. The Performance monitoring tool itself, as a running software, also consumes system resources, therefore, when choosing the monitoring tools, we should choose the tools that have less impact on the system, and set the appropriate monitoring level and sampling period to ensure that the operation of the application system is not affected, and sufficient data can be provided for analyzing the performance problem. The performance monitoring of WebSphere Commerce E-commerce Web site can be divided into: Operating system monitoring, WEB server monitoring, WebSphere Application Server monitoring and database server monitoring, and so on.
Operating system Monitoring
Operating system resources are the foundation of the operation of various application software, the lack of system resources will seriously affect the performance of the application software, during the operation of the production system should always pay attention to the use of system resources. When the application is running, it usually uses CPU, memory and I/O resources, and these three resources are also the focus of operating system resource monitoring.
The main function of CPU is to interpret computer instructions and to handle data in computer software. The CPU has 4 running states: System, User, wait, and Idle. System state refers to the CPU in the execution of the operating system kernel code, the User state is the CPU in the execution of program code, wait state when the CPU waiting for I/O operations, and Idle state when the CPU is idle. When the e-commerce Web site is running, the total CPU utilization of the server should not exceed 50%~70%, if the overall CPU usage is too high, the user experience of the page opening time may be affected. In a variety of running states, the normal time should be the execution of program code of the User state accounted for relatively high. If the system state occupancy rate is high, the process is busy, and if the wait occupancy rate is too high, the system is often in I/O waiting state, both of which indicate that a system may have some sort of exception.
Memory is used to store the CPU processing of temporary data and program instructions, buffering and data exchange functions. Generally, both the application server and the database server should be equipped with sufficient physical memory and the maximum memory usage parameters of the JAVA virtual machine and the database server, combined with the size of the physical disk, to avoid performance degradation due to the use of virtual memory space on the disk. It is to be noted that the operating system uses part of the physical memory space for the file cache, so it's not easy to think that memory usage is a problem with a memory footprint of nearly 100%, but it should focus on the exchange of physical and virtual memory, indicating a possible performance problem if the frequency of the exchange is too high.
I/O is the system input and output, including the application of disk reading and writing, network transmission and other peripherals to read and write. Disk I/O reflects the behavior of the server reading and writing disk data, logging, and so on. Network I/O reflects the behavior of receiving requests, providing responses, and reading and writing network resources, such as remote databases. I/O monitoring should also focus on the amount of I/O data and the number of I/OS, under the same size of I/O data, too frequent I/o times, can also lead to the exhaustion of I/O resources, become the bottleneck of the system. Disk busy is an important indicator of concern when monitoring disc I/O.
Nmon is a tool that IBM provides to monitor and analyze AIX and Linux operating system resources. After the V6 version of the AIX operating system has been installed by default Nmon tools, the recommended operating system with the NMON monitoring operating system resources to use. In addition, IBM DeveloperWorks provides a variety of operating system Nmon tools, Nmon manuals and Nmon_analyser analysis tools and other rich resources to choose from.
Compared with other operating system resource monitoring tools, Nmon records the information is very comprehensive, and provides in the interactive mode and data collection two monitoring modes, interactive mode is in the Nmon window dynamically observe the change of monitoring information, commonly used in problem diagnosis; data The collection mode is to write all the monitoring information into the log file, and then use the Nmon_analyser analysis later, suitable for daily system monitoring and historical trend analysis. Nmon can monitor the use of system resources mainly include the following data:
CPU occupancy Rate
Memory usage
Kernel statistics and running queue information
Disk I/O speed, transfer and read/write ratios
Utilization of File system
Disk Adapter Usage
Network I/O speed, transmission and read/write ratios, error statistics, network transport packet size
Page space and page I/O speed
Processes that consume the most resources
User-defined Disk groups
Computer details and resources
asynchronous I/O, available only for AIX
Workload Manager (WLM), available only for AIX
Network File System (NFS)