What's New
The new capabilities in Sun Grid Engine 6.2 include dramatic scalability improvements and enhancements to usability and management. Together they enable you to get even higher utilization from your compute resources.
Improved Scalability
Sun Grid Engine 6.2 supports compute clusters with up to 63,000 cores, processing parallel jobs with tens of thousands of tasks. This unsurpassed scalability comes from improved interactive job support, and streamlining communications between components.
Advance Reservation
The Advance Reservation feature lets users schedule compute jobs, and the Sun Grid Engine 6.2 matches reservations with resources to maximize utilization.
Array Task Interdependencies
This community-contributed feature evolves the traditional all-or-nothing structure of job dependencies to Sun Grid Engine 6.2's system where dependencies can be "chunked"—one task can depend on several, or several tasks can depend on one, creating a more sophisticated and faster workflow.
Multi-Cluster Resource Sharing
The Sun Grid Engine 6.2 Service Domain Manager allows two or more clusters to share resources intelligently. Clusters are monitored according to service level objectives, and policies can control relative cluster priorities.
Why a Cluster?
These diagrams show you how a compute cluster works, and Sun Grid Engine can make more efficient use of your computers.
Improving Utilization with Sun Grid Engine
A typical computing environment operates at approximately 10-25% utilization. Usage policies are ad hoc, and accounting and reporting functions are rudimentary if they exist at all. This type of environment creates management and procurement headaches, not to mention the waste of resources and money.
Sun Grid Engine 6.2 consolidates resources by creating compute clusters, which distribute the overall system load and boost utilization rates up to 95% and more. Grid Engine also provides a way to create and enforce usage policies. Management, procurement, and accounting are simpler and more predictable.
|
Listen and Learn
Listen as Sun Radio News host Russ Castronovo talks with chief architect Daniel Templeton about the new features of Sun Grid Engine 6.2, and how it can make your compute cluster more efficient.
Fritz Ferst, director of Sun Grid Engine, and Miha Ahronivitz, Sun Grid Engine product line manager, talk about what value Sun Grid Engine brings to customers.
Get Sun Grid Engine 6.2 Today:
Sun Grid Engine in Action
Sun Grid Engine is the world's most popular open source DRM system, with more than 10,000 installations worldwide. At the Tokyo Institute of Technology, Sun Grid Engine helps power the largest supercomputer in Asia and the 16th largest in the world. This 655-node supercomputer runs on Sun hardware as well, including Sun Fire x4600 servers.
The Ranger supercomputer at the Texas Advanced Computing Center (TACC) is the fourth largest in the world and Sun Grid Engine helps keep it humming.Tom Minyard, TACC Assistant Director, explains why Sun helped build the largest general-purpose supercomputer in the world.
Sun Grid Engine in the Field
- Callidus Software uses the Sun Grid Engine to efficiently provide high-performance on-demand business computing solutions to its customers. Read More about Callidus
- Sun Grid Compute Utility provides a scalable, cost-effective infrastructure for Virtual Compute Corporation's on-demand, high-performance computing infrastructure to run its customers' compute-intensive scientific and engineering applications Virtual Compute Corporation
- Sun Grid Engine manages AMD's internal grid of more than 5,000 servers, 500 workstations and well over 10,000 CPUs, deployed across six design centers worldwide that support all phases of microelectronic design, circuit analysis, integration, electrical characterization and mask generation. Learn more about AMD
|
