Cluster Management

Cluster management can be split into two categories:

1. Deployment and Ongoing Management/Maintenance,
2. Application Management.

There are several tools available to support cluster deployment. Our recommended solution is System Imager. Upon request, we can install and configure System Imager as part of a complete cluster solution. System Imager works along the client-server concept. An image server stores the required image, a snapshot of an entire installation and configuration details such as network address and hostname, which can then be replicated across a network to any number of compute nodes. Multiple images can be stored on the image server and farmed out to any number of systems in accordance to a pre-determined criteria such as whether a node will support a particular piece of hardware. All the replication can be done remotely in a totally hands-off environment. Furthermore, when an image is updated, perhaps to add a software package or to update an existing package, only the change is synchronized to the clients as opposed to the entire image.

On the application management side there are also several tools available to support parallel execution of jobs. Sun™ Grid Engine and OpenPBS are two options that are available from Xinit as part of our Cluster Consulting & Integration Services. Both solutions can be applied to manage applications that require parallel execution on multiple nodes.