Cluster Resources has launched Moab Workload Manager 5.2, which is the latest version of its management tool to support hybrid clusters, job arrays and hierarchical fairshare.
Moab Workload Manager is a policy-based job scheduler and event engine that enables utility-based computing for clusters. Moab Workload Manager combines intelligent scheduling of resources with advanced reservations to process jobs on the right resources at the right time.
‘Moab Workload Manager 5.2 contains numerous new and upgraded features that will enhance user experience and increase the efficiency of a cluster or data center,’ said Michael Jackson, president of Cluster Resources. ‘The most notable of these enhancements is support for hybrid clusters, which allows us to provide a cost-efficient solution for our customers running multiple operating systems.'
Running one Linux cluster and one Windows cluster, each with different peak usage and idle times is expensive and inefficient; running Windows on a Linux cluster yields maximum hardware utilization and ROI.
Moab 5.2 removes the barrier of a node's operating system by maximising workload throughput with a single self-optimising system that balances the number of nodes running a particular OS against user/group and workload service levels. When conditions are met, Moab triggers the change via a site's preferred OS-modification technology, such as diskfull and diskless provisioning, dual boot or virtualisation (i.e. via Hyper-V within Windows Server 2008, VMware, or Xen).
The Moab hybrid cluster model also consolidates administration and centralises job submission across both platforms. Administrators can manage both operating systems' policies and workload from Moab's unified console.
Moab also makes the dual-OS nature of the cluster transparent to end users by applying application and workload information that ensures jobs run on the correct OS without the user needing to specify it.
This latest release of Moab also has improved beta support for the Sun Grid Engine (SGE) resource management system. Moab is able to externally schedule SGE to manage resources according to workload needs and policies defined in Moab.
Additional new or enhanced features in 5.2 include hierarchical fairshare, job arrays, job templates, and the industry standard 'Autoconf' configuration script to speed and dynamically customize Moab installations.
Enhancements have been made to utility and data centre computing, and hosting centre applications. Moab's high throughput mode for utility computing allows behind the scenes optimisation in job submission, job migration, job start, and job tracking, enabling job throughput rates of up to 100 jobs per second in clusters running both Moab and TORQUE resource manager. Dynamic partitions enable data centres to manage workflow peaks by optimising the available resources across web farm pools.
Additionally, improvements to dynamic jobs allow workloads to flow around hardware failures and maintenance stops. Hosting centre improvements include extended and refined trigger functionality and additional support to application-based resource managers.