Moab Workload Manager for Grids^TM (Cluster Scheduler & Policy Engine)

Overview:

While Moab is significantly more powerful than previous generation products, it also possesses a far more intuitive interface allowing sites to more quickly take full advantage of their cluster's potential.  There are four main categories of focus for Moab.  First, a next generation policy engine, second the addition of multi-resource management support, third is a significantly improved application integration ability and lastly simplified interface access.

Next Generation Policy Engine:

Built upon a dynamic execution framework, the next generation policy engine includes scheduler and resource policies, event policies and with the Moab Grid Manager & Scheduler the addition of Grid policies.

• Scheduling Optimizations - allow intelligent schedule re-ordering and resource allocation to maximize cluster performance and responsiveness
• Advance Reservations - reserve any set of resources dynamically for use by any group for any period of time
• Malleable Jobs - set policies allowing Moab to modify existing jobs to maximize system utilization and minimize end user response time
• Integrated Diagnostic Services - determine why a job is unable to run and determine the full history of a job months after it completed. Determine which nodes are failing and how often or how well your resource manager is performing under the current load
• Object Messages - associate informative messages regarding object failures and history directly to the affected object allowing near immediate analysis of cluster issues
• Job Preemption - Allow preemption of active jobs to accomodate higher priority workload and enable management of desktop resources
• Fairness Policy - Allow fine grained control of cycle distribution over varying time scales through flexible prioritization, throttling policies, fairshare, allocation management, and QoS services
• Event Policies/Triggers - allow arbitrary actions to be associated with cluster events
• Grid Integration - Track grid workload and allow control over the level of service and quant ity of resources made available to remotely submitted workload

Multi-Resource Management Support:

• Multi-Resource Manager Support - allow Moab to query multiple sources of information to obtain a full-picture view of available resources
• Generic Resource Support - schedule compute nodes as well as node-locked and floating cluster resources such as filesystems, licenses, or tape-drives

Application Integration:

• Identity Management - allow the scheduler to interact with identity and security credentials from Globus or other sources
• Resource Provisioning - allow the scheduler to dynamically modify node resources by installing software packages or patches and operating system to meet current workload needs through integration with software provisioning tools such as Red Carpet Enterprise or other provisioning tools
• Allocation/Billing Manager - allow the scheduler to interact with allocation/billing managers to provide utility like billing reports or managed credit reports using tools such as QBank, Gold or others
• Databases - allow the scheduler to interact with databases to either provide or consume information that is used/provided by another application or data store

Interface Access:

• Powerful Graphical Management Tools - truly visual your cluster including its current state, historical performance, and anticipated future behavior as well as manage it with empowering wizards and various automated capabilities through Moab Workload Manager
• Moab Access Portal - allow users to submit and manage their own workload through a simple web-based portal so as to maximize accessibility and minimize client dependencies (in development)
• Vendor Portal - allow user access via your existing vendor portal using standard command calls and professional integration assistance from Cluster Resources
• Custom Portal - allow user access via your own custom portal using standard conmand calls and professional integration assistance from Cluster Resources

Moab's design is both powerful and intelligently integrated allowing policies to be highly optimized.   It is acknowledged by many as "the most advanced scheduler in the world" and continues to outperform all competitors in cluster performance benchmarks and shoot-outs.  It is currently in use at over a thousand of the world's leading government, academic, and commercial sites significantly improving manageability and efficiency of systems ranging in size from one to several thousand processors.

Additional features include

• Extensive resource utilization tracking and statistics
• Non-intrusive 'Test' modes
• Advanced built-in HPC simulator for analyzing workload, resource, and policy changes

PBS Scheduling API - TORQUE, OpenPBS and PBSPro
Loadleveler Scheduling API - Loadleveler (IBM)
SGE Scheduling API - Sun Grid Engine (Sun)*
BProc Scheduling API - BProc (Scyld)**
SSS XML Scheduling API*
LSF Scheduling API - LSF (Platform)
Wiki FlatText Scheduling API (Wiki)
*partial support or under development
**supported under Clubmask

Moab is currently supported on all known variants of Linux, AIX, OSF/Tru-64, Solaris, HP-UX, IRIX, FreeBSD, and other UNIX platforms.

Evaluating Moab

Moab is available to customers and business partners with evaluation releases available to a broader audience.  Please contact us to learn more about what Moab can do for you.