The title Sun Grid Engine 6.2 for Dummies clearly would not have worked, since natural selection alone excludes that segment of any possible audience, so Daniel Templeton prudently titled his white paper
Beginner\'s Guide to Sun Grid Engine 6.2: Installation and Configuration. The author's credentials could not be more authoritative. Templeton is Strategic Liaison Manager for the Sun Grid Engine product team.
Templeton walks his readers through basic installation and configuration of Sun Grid Engine 6.2, explains Sun Grid Engine architectural concepts, and presents some suggested configurations for clusters of various sizes. The target audience is system administrators who have downloaded the Sun Grid Engine 6.2 evaluation version or the Grid Engine 6.2 courtesy binaries or who have just purchased the Sun Grid Engine 6.2 software.
The author points out that the Sun Grid Engine software is based on the Grid Engine open source project, which has a thriving community of users and developers who are working with the software on a daily basis. For questions about installation and configuration not covered in the Sun Grid Engine documentation, the Grid Engine open source community can be an invaluable resource. The community mailing lists, wiki, blogs, and how-tos cover a broad range of useful topics. See the Grid Engine open source project site for more information.
The Sun Grid Engine software, Templeton continues, is a very flexible tool for increasing resource utilization and overall job throughput while managing resource usage according to business policies and goals. It is capable of scaling from the smallest departmental clusters all the way up to the enormous Ranger cluster at the Texas Advanced Computing Center, which boasts more than sixty-two thousand CPU cores and over one hundred terabytes of aggregate memory.
A distributed resource management (DRM) system, the Sun Grid Engine software enables higher utilization, better workload throughput, and higher end-user productivity from existing compute resources, Templeton writes. By transparently selecting the resources that are best suited for each segment of work, the Sun Grid Engine software is able to distribute the workload efficiently across the resource pool, even as it shields end users from the inner working of the compute cluster.
The Sun Grid Engine software gives administrators the ability to accurately model as a resource any aspect of the compute environment that can be measured, calculated, specified, or derived, facilitating the most effective and efficient scheduling of workload to available resources. The Sun Grid Engine software can monitor and manage resources that are concrete, such as CPU cores or system memory, as well as abstract resources, like application licenses or mounted file systems, the author notes.
Not just a tool for allocating resources based on workload requirements, the Sun Grid Engine software also provides numerous advanced scheduling features that allow an administrator to model not only the compute environment, but also the business rules that control how the resources in that compute environment should be allocated. Some of the advanced scheduling features offered by the Sun Grid Engine 6.2 software that Templeton enumerates are advance reservation of resources, fair-share scheduling, scheduling by resource value, observation of deadline times for specific segments of work, user-specified relative priorities, and starvation prevention for resource-hungry workload segments.
An overview of Templeton's white paper follows:
- Chapter 1: "Introduction to Sun Grid Engine," describes some basic Sun Grid Engine concepts.
- Chapter 2: "Sun Grid Engine Scheduler," introduces the complete list of scheduling policies and what problems each one solves.
- Chapter 3: "Planning a Sun Grid Engine Installation," provides suggestions for preparing the computing environment for a cluster installation.
- Chapter 4: "Suggested Sun Grid Engine Configurations," presents some common cluster types with suggested configuration options.
Beginner\'s Guide to Sun Grid Engine 6.2: Installation and Configuration.
[...read more...]