| This post was kindly contributed by Key Happenings at support.sas.com - go there to comment and to read the full post. |
Contributed by Bill Nasuti, Global Professional Services, SAS
We often hear folks refer to SAS Grid as if it were a single entity, when in reality it consists of many facets. SAS Grid provides a rich functional toolset to improve performance and resource utilization of SAS applications and programs. Let’s dig a little deeper and look at the different ways SAS Grid can help us improve performance and improve resource utilization.
SAS Grid Computing involves employing multiple computers collectively to improve hardware utilization and reduce processing times; this is often times referenced under the HPC or High Performance Computing umbrella. SAS Grid computing is built upon Platform LSF which enables our ability to manage workloads across multiple platforms. By employing a third party product like Platform LSF, we can decouple SAS from the underlying OS. This provides us the opportunity to more efficiently manage SAS application workloads for many virtual and physical platforms.
Let’s take a look at how we can utilize this exciting technology with SAS.
- Multi-User Workload Balancing – By submitting jobs to the grid rather than an individual machine we can spread the workload out among all the machines that comprise the grid. This provides us a plethora of enterprise class IT governance. For example, we can ensure that machines are not overloaded and workloads are distributed based upon resource requirements. We can also assign priorities based on users or applications and even suspend or preempt jobs. Multi-User workload balancing allows SAS the ability to ensure that their customers are fully utilizing their SAS resources.
- Parallelized Workload Balancing – This adds a whole new dimension to SAS processing, why run a job serially when you can now break it up into smaller pieces that can be run in parallel? Dramatic performance gains can be achieved utilizing this technology. In some cases, programs that used to take days can be run in minutes.
- Distributed Enterprise Scheduling – Some jobs have to be run on a regular basis and are very computational intensive. We typically run these jobs on the batch server and schedule them to run off hours when there is less interactive load for the SAS resources. By integrating enterprise scheduling with the other grid technologies, we can ensure that customers are efficiently utilizing their SAS resources even during off hours.
- High Availability – By employing a load balancing switch or corporate DNS, SAS grid will provide a high availability platform for many SAS servers. You can now easily configure a primary and failover metadata server, remote services, or many other SAS servers.
- Scalability/Maintainability – SAS grid platforms on UNIX provide unique opportunities by allowing us the ability to share configuration and binary directories. This dramatically simplifies our ability to scale across the grid by allowing us to very simply add grid nodes. Maintainability is simplified as we can very easily add and remove nodes on the grid for maintenance.
In summary, SAS Grid Manager is an exciting technology that will afford SAS customers the ability to take full advantage of their SAS resources across the enterprise. In the future we will take a more in depth look into how SAS grid works. In the mean time, if you would like to learn more regarding how to deploy grid, here are some useful links:
| This post was kindly contributed by Key Happenings at support.sas.com - go there to comment and to read the full post. |