Newsletter - February/March 2007
Optimizing your Compute Resources
Cluster Resources, Inc. recently unveiled a new partnership with Panasas, Inc., delivering a holistic next-generation, compute and storage solution through the integration of Cluster Resources’ Moab Workload ManagerTM with the Panasas ActiveScale® Storage Cluster. This combined solution is believed by many to be the next logical step in the evolution of the cluster, allowing more unified system optimization, more adaptable resources and more powerful policy-based service delivery.
The integration of these technologies provides with the information it needs to more efficiently
drive user workloads in a storage-aware manner, avoiding failures and data bottlenecks while intelligently directing the ActiveScale Storage Cluster to adapt to storage requirements of the upcoming workload schedule. This partnership is creating a complete clustered environment
that optimizes compute and storage resources for maximized return on investment.
“The collaborative work of Cluster Resources and Panasas enables HPC customers to more fully maximize the value of their infrastructure by tightly mapping Panasas Storage Clusters to the workload that accesses them under a single policy engine,” said Len Rosenthal, chief marketing officer of Panasas. “Customers will have the ability to schedule and run workloads
in order to achieve maximum I/O performance in an intelligent and highly adaptive manner in line with high-level mission objectives.”
Cluster Resources and Panasas are currently delivering on the early phases of the integration effort allowing to automatically learn and even predict the I/O requirements and characteristics of cluster workload. Further, Panasas is generating and exporting storage contention information that is now being used by to prevent over-commitment of storage resources, minimizing data thrashing through intelligent workload selection.
Read the full press release here
The time is getting closer to Cluster Resource’s free technical training on March 23 and 26. Slots are still available for both days. The training will consist of six one-hour classes repeated two days. Customers and partners will be given first priority and other slots will be filled as space allows. At the same time, we are hosting on site training for anyone that can travel to our Provo, UT headquarters location. This will allow you to meet face-to-face with the developers.
The six topics to be covered are:
Prioritization and Fairshare
in the data center
Complete details about connecting into the conference, call in numbers, training order, etc. will be provided next week to those that sign up or inquire about the training.
If you would like to receive an invitation to the training, please e-mail email@example.com.
Training recordings will be posted on line after the live training session. To view last year’s training session, go here.
5.0 evaluation builds are available on the Cluster Resources eval Web site. When 5.1 is released soon, an updated evaluation build will be available as well. Evaluations are fully functional and also allow you to try out the graphical tools, Moab Cluster Manager and Moab Access Portal. Evaluations last for 45 days before you will need to contact Cluster Resources to obtain a license.
The new releases of are significant because they mark definitive upgrades to their grid and utility computing abilities. 5.0 has added functionality in the areas of data staging, virtualization, new diagnostics, and scalability. 5.1 will improve upon that even more with added integration with SLURM. More information will be available soon regarding the 5.1 release. See the CHANGES file contained in the binary for a detailed explanation of new features, enhancements, and fixes.
These new evaluation builds give you the opportunity to try many of the new features included in 5.x.
To download the evaluation for clusters go here
To download the evaluation for grids go here
The admin manual PDF has been updated with changes for 5.0. This PDF will be updated again soon with the release of Moab 5.1.
You can view the PDF here
(The link is next to the Admin Manual for Moab Workload Manager)
Reservations contain multiple ACL modifiers to help change the default behavior of ACL processing.
The recently added ACL modifier, HPEnable, allows you to determine which credentials receive preferential access to a reservation. By adding a ~ to a credential list, such as USERLIST, the credential will be ignored during soft policy scheduling and will only be considered for hard-policy scheduling once all eligible soft policy jobs have been started.
In this example, all John's jobs will allowed to run in the reservation at any time. Bill’s jobs will also be allowed to run in this reservation, but only considered after all of John's jobs are given an opportunity to start. John's jobs will be considered before Bill’s jobs regardless of job priority.
For more information on HPEnable see 22.214.171.124.5 ACL Modifiers in the documentation.
Product Feedback, Technical Support, Consulting and Custom Development queries can be sent to firstname.lastname@example.org or call us at +1 (801) 717-3700
Moab Workload ManagerTM, Moab Cluster ManagerTM and MonitorTM, and Moab Access PortalTM are trademarks or registered trademarks of Cluster Resources IncorporatedTM.