IBM Workplace Server Performance and Health Monitor
A Web application that enables users to gauge the performance and health of a Workplace server.
Date Posted: June 22, 2006
|
|
 |
|
|
|
Follow these instructions:
- First, you will want to establish a response time baseline for your current workload. After you have that baseline, you will be able to quantify the effect of any changes you make to the system environment. This can include making performance tuning adjustments, as well as adding new users or applications to your server.
- When the monitor detects a failing Workplace component, it can send you an alert via e-mail. The e-mail message will include a description of the error, along with a time stamp. This information allows you to immediately start isolating and analyzing the problem.
| | |
|
The IBM Workplace Performance and Health Monitor is the only tool that lets you view the performance of a Workplace server at a level of detail that is similar to that experienced by your users. For instance, it will report how long it takes to create a document folder, how long it takes to post a new discussion topic, and how long it takes to refresh the mail inbox. Other tools, such as Tivoli® Performance Viewer, focus on the lower-level resources, such as JVM memory usage and JDBC connection pool queues, within the application server. | | |
|
No. An installation script is provided for single-server i5/OS, Windows, and Linux deployments. After you bring up the monitor Web application in your browser, you can launch a monitor collection with a single mouse click. At that point, the monitor runs completely unattended. It even restarts itself after a server outage. | | |
|
The monitor's user interface maintains the most recently-collected data and provides a viewer for it. For historical data, the monitor maintains a set of HTML files from which the results can be viewed. The amount of historical data retained by the monitor is configurable such that it will automatically clean up and delete expired log files. | | |
|
In addition to logging data to viewable HTML files, you can also log response time data in comma-separated value (CSV) formatted files. Data in this format can be imported directly into many applications. | | |
|
Yes, although the installation script only works for single-server deployments. The "Installation and User's Guide" includes instructions on how to install the monitor in a network deployment. After being installed and activated, the monitor runs independently on each node in the network deployment. | | |
|
Since the monitor runs on the application server that it is monitoring, it will stop running when its hosting server terminates. But before terminating, it will send an e-mail notification to warn you that the server is ending. When the server comes back up, it will send another e-mail notification to tell you that the server has restarted. And, finally, it resumes its data collection where it left off before the server ended. | | |
|
No; the monitor is simply an application that uses externally available APIs that are included in the Workplace Toolkit to drive a workload. It cannot see what is going on inside the Workplace server itself and it has no awareness of other actual users, other than the effects that other activity has on the response times of the transactions that it generates. | | |
|
Below is a list of best practices:
- Register a special Workplace user under which the Workplace transactions will be run. In the process of running the Workplace transactions, objects such as calendar entries, mail messages, documents, and discussion forums will temporarily become visible if the actual user signs in to Workplace. This could create confusion and could even disrupt the monitor if the user were to tamper with those objects. The simplest solution is to register a special user and specify that user's ID and password on the monitor configuration panel.
- Ensure that your Workplace server is functional before starting the monitor. Sign in to Workplace under the special user ID and manually verify that the components you plan to monitor are working as expected. For instance, send an e-mail message to yourself, create a document library, create a discussion team space, create a new calendar entry, and create a new Web conference. These actions are especially important if you are going to monitor the messaging component, because it will ensure that the user's mailbox has been provisioned.
- Import the response time data into analysis tools to reveal response time trends and spikes. For instance, most spreadsheet applications can directly open the CSV-formatted log files and easily generate graphs of the Workplace server's response time.
| | |
|
Workplace Collaboration Services 2.6.x
Workplace Services Express 2.6.x | |
|
|
|
|
For platform(s):
Linux - Red Hat, Linux - SuSE, OS/400, Red Hat Linux, Windows 2000 Advanced Server, Windows 2000 Server, iSeries
|
 |
For topics:
J2EE, monitoring, performance, testing, Web service, Servers, Workplace
|
|
| |
