IBM Cluster Monitor
A one-stop, easy-to-use, configurable, fully automated, Web-based tool for monitoring and observing cluster-wide performance in the form of visible graphs.
Date Posted: May 6, 2008
|
|
 |
|
|
What is IBM Cluster Monitor?
Clustering systems provide commands for obtaining statistics; however, the results must be compiled and presented either manually or with the help of script. Often there are problems with lack of graphical output, accuracy, portability, and configurability. Much effort is required in developing such scripts. With IBM® Cluster Monitor, you can define statistics collection and customize them for the monitoring of cluster performance, health, network load distribution, etc.
IBM Cluster Monitor provides the following:
- an easy-to-use interface for creating code for statistic collection, including infrastructure for reporting statistics
- configuration of cluster on machines on which statistic collection is to be deployed
- configuration of a Web view through a Web interface, including the ability to change settings, add or modify graphs to be generated, configure views, and deploy statistic collection.
The reduced effort and time allows anyone to develop code for the monitoring a group of machines for performance analysis and to use such code to demonstrate efficiency, performance values, or simple statistics. The tool enables configuration of the monitoring system at the click of a button. Administrators can create desired views and statistics to be displayed using a browser. Performance statistics are displayed in the form of moving graphs. Configuration settings enable a user to define his own view of the performance; a view can contain one or more graphs. Graphs are generated based on a template, which can be customized for the user's need.
How does it work?
IBM Cluster Monitor uses a tool called ganglia as a back end for collecting and consolidating various statistics to be displayed through a browser. Ganglia contains two packages, gmond and gmetad, for collecting and consolidating metrics, respectively. Another tool called rrd generates graphs in the form of images to be displayed on top of script.
A set of predefined, easy hook functions must be implemented for a third-party cluster and must be monitored with IBM Cluster Monitor. Hook functions collect statistics. Example hook functions have been written and included for IBM General Parallel File System (GPFS) and Network File System (NFS).
|
|
|
| | About the technology author(s):
 Ujjwal Lanjewar joined IBM in 2000 and works as a senior staff software engineer with the storage and file systems team in Pune, India. For the past years, he has been involved in the development and support of IBM distributed file systems products. Mr. Lanjewar's major contributions include work on the Andrew File System and the SAN File System on the HP-UX 11i platform. In 1997, he received an M.S. in software systems from BITS Pilani, India.
|
 Ashwini Athalye joined IBM in 2006 and worked until 2007 as an associate software engineer on federated file system and GPFS (general parallel file system). Her expertise includes storage and file systems. She received her B.E. in computer science from Cummins College of Engineering Pune, India, and is currently working on her M.S. in computer science.
|
IBM is a trademark of IBM Corporation in the United States, other countries, or both.
Other company, product, or service names may be trademarks or service marks of others.
|
|
|
| |
