|
Flexible Resource Quotas
|
Allows limiting the maximum number of running jobs per
user, user group, and projects on arbitrary resources like queues,
hosts, memory, software licenses.
|
Allows flexible allocation of grid resources according to policies.
|
|
Boolean Expressions
|
Allows a user to specify a job's resource requirements using simple compound logical expressions. For example, a user could request that a job runs on a host which fulfills the condition "Solaris or Linux but not Linux on IA64".
|
Ability to fine tune the grid resource to the most important projects in real time.
|
|
Latest Standard Applications API (DRMAA) 1.0
|
Distributed Resource Management Application API (DRMAA) is a set of standard APIs developed by the Global Grid Forum for application builders, portal builders and ISV's. Release 6.1 supports the latest C and Java bindings of DRMAA 1.0. Additionally, the following are provided for backward compatibility: DRMAA 0.5 Java binding and DRMAA 0.95 C binding.
|
ISV's can easily integrate their applications with Sun Grid Engine 6.1 without having to learn the internals of Sun Grid Engine.
|
|
Sun Grid Engine 6.1 Accounting and Reporting Console
|
Sun Grid Engine 6.1 saves job accounting information into a relational database (Supported Databases are Oracle, MySQL and PostgreSQL - see support matrix).
Users can generate usage reports, e.g. Accounting per department/project/user, number of jobs completed etc.
|
With the Accounting and Reporting Console it is easy to determine how compute resources are being used. Data can sent to an external billing module for invoicing, cross-charging etc.
|
|
Heterogeneous platform support
|
Sun Grid Engine 6.1 supports all major operating systems and platforms. See the support matrix.
|
Maximizes flexibility and usage of IT assets.
|
|
Cluster Queues
|
A queue may span more than one execution host.
|
Allows managing thousands of hosts with a few simple commands.
|
|
Resource Reservations and Backfilling
|
The grid is filled with smaller jobs while the large jobs are being scheduled to execute.
|
Allows large, compute-intensive jobs to automatically be scheduled and run without sacrificing utilization.
|
|
Array jobs
|
Provides repeated execution of the same (or very similar) set of operations, for example during graphical rendering of multiple frames.
|
Maximizes efficiency of designers and engineers using the grid.
|
|
Parallel environment (PE) management
|
Manages parallel applications (MPI or PVM enabled) through a dedicated interface.
|
Allow rapid completion of demanding design and financial computations using parallel processing.
|
|
Policy-based resource allocation
|
Allows distribution of resources to teams or departments, e.g. proportionally to their financial contribution.
|
Allocates compute power according to business policies.
|
