Today, the onslaught of data and the corresponding need to process it quickly has created significant demands on traditional IT infrastructure. To meet that need, high-performance computing (HPC) has become an integral part of the workload mix. To complicate matters, IT infrastructure is in transition as companies move towards hybrid processing models which include traditional bare metal, virtual machines, cloud and containers.
This complex infrastructure cannot function efficiently using simplistic workload scheduling tools. Each type of processing model requires a slightly different approach to achieve the highest level of efficiency, therefore creating potentially conflicting operational focus which a simple workload scheduling tool cannot satisfy.
A greater focus is needed on workload management tools to simplify operations, balance resources, manage job priorities, and performance.
Simplified operations are achieved through workflow automation, easy to use GUI, intuitive reporting dashboards, automated alerts, and customization.
Another major challenge with HPC is the high cost of software licensing. In some cases, the cost of the software can be twice as much as the cost of the HPC infrastructure. A platform that makes prudent use of software licenses through intelligent scheduling can help to significantly reduce costs.
The main premise of this paper is to illustrate how a software-defined infrastructure (SDI) platform that goes beyond traditional scheduling tools will reduce the overall cost of HPC. In a
traditional sense, most customers attempt to procure as much hardware as their budget will allow. To make up for the lack of budget, many customers migrate towards open-source
solutions as the general perception is that these are less costly. The reality is that in many cases, the cost of support more than offsets the software license costs.
In this document, we will introduce the concept of “infrastructure efficiency”; with a platform that not only manages complex workloads but does this in such a way that is easy to operate and monitor thus maximizing resource pools and reducing cost.
We look at an example of the total cost of ownership (TCO) of the IBM Spectrum LSF Suite for Enterprise compared to a pure open-source alternative. While there is a lot of controversy associated with the use of open-source, this will not be covered in this paper. Edison’s view is that while unmanaged use of open-source is high-risk, open-source code integrated with
proprietary offerings and managed by a legitimate technology service organization is a good business practice for both provider and end-user as it provides a way to lower cost and maintain low risk levels.
For IBM’s perspective on HPC written by Gabor Samu, Portfolio Marketing Manager/High-Performance Computing, click here.
Click here for full paper