In addition to our hardware-based in-house developments, with ClustWare® we offer a comprehensive management and monitoring tool. This combined solution, comprising hardware and software modules, allows you to centrally control the cluster system and read specific performance parameters.
For the use of ClustWare®, an appliance is required via which the central management service and a database server are provided. The system daemons, which collect operating parameters and forward these to the database, run on the individual compute nodes. The information collected is prepared as a graph via a Web GUI and displayed to the user.
In addition to system monitoring, the software also allows the user to set threshold values. These upper limits can be configured for various parameters, such as system temperature and load. Where any threshold parameters are exceeded, notification of this will be automatically sent by e-mail.
In addition to monitoring and early fault detection, ClustWare® offers a variety of interfaces for connection to the Job Scheduler. The combination of management software and Job Scheduler creates the necessary basis for advanced functions, particularly for realizing energy saving potential. By means of reading the current overall load, individual idle nodes can be shut down. If the load increases, any compute nodes that have been shut down will be restarted, and thereby provide additional computing power.