|
|
|
|
Grid Computing: Get More Mileage from Your IT Resources
|
The concept of grid computing is now being embraced by a growing number of organisations, and the reasons are clear. First, IT strategies often focus on reducing costs, and grid computing is a far less expensive alternative to purchasing new, larger server platforms for specific applications and projects. You can do more with your existing, possibly underutilised hardware. Second, many organisations process such large data volumes and perform repetitive tasks to the extent that workload requirements exceed server capabilities. In such environments, grid computing is a way of applying resource from a collection of computers in your network and channelling this power towards a single project.
The net result is to leverage your IT infrastructure to optimise all compute resources, to manage your data and workloads in more efficient ways, and deliver faster results. The good news? MP CONNECT (multi-processing) functionality, included in SAS since Version 8, enables grid computing. And there's more to come.
Choosing the right applications
It's important to identify applications that complement a grid approach, so realistic expectations can be set with achievable performance goals. Good candidates are applications that: take many hours and possibly even days or weeks to run; involve many repeat runs of a task; process large amounts of data; and decompose into execution units, data subsets or both. So, once you've identified your applications, how can SAS help?
SAS on the grid: firing on all cylinders
In 2003 SAS became the first enterprise BI vendor to join the Global Grid Forum, a group of over 5,000 researchers and practitioners working on distributed computing and grid technologies. However, the concepts behind grid computing weren't new to SAS: indeed, Version 8 introduced MP CONNECT functionality to segment a job workload into independent units of work, which are then processed in parallel across any number of heterogeneous computers within a network.
Since then, SAS has gone much further to deliver more integrated and complete grid capabilities across SASŪ9. And later in 2005, we plan to announce further integration between SAS and Platform Computing to provide more configurable dynamic resource-based load balancing capabilities, as well as grid monitoring. Indeed, advantages provided by the Platform Computing Job Scheduler cover critical areas such as automation, reliability, scalability and prioritising workload.
The grid in action
So are SAS users deploying this technology to gain competitive advantage? They certainly are. Here's one example from the US. According to the Director of the Center for Advanced Analytics and BI at Texas Tech University, "Research is an ongoing process. The ability to move and adapt to change quickly is appealing. Our SAS grid allows me to go in a lot of different directions and to do these very quickly." You can find more details at www.sas.com/success/texastech.html
In summary, SAS on the grid does enable you to do more with less; to achieve higher performance and, ultimately, lower cost IT ownership. You can optimise unused capacity in a network, handle exploding workload requirements and deliver essential business information in a faster time frame - all while staying within your IT budgets.
For more information and to download a white paper on Grid Computing and SAS visit: http://support.sas.com/rnd/scalability
Please visit www.ggf.org for more details on the Global Grid Forum
|
