Administering a high performance technical computing environment can be a challenging undertaking. There are many facets to consider, including efficiency, reliability and application performance. In technical computing environments, you’ll often find a variety of workloads running, including in-house codes, open-source and commercial, licensed applications. Special consideration is required for commercial, licensed applications to ensure that these finite resources are allocated, and shared optimally according to business or project demands. On the surface, managing software license resources may seem intuitive, but like the proverbial tip of the iceberg there is much more to consider than meets the eye. As you will see, managing license tokens is not trivial.
“EDA software tools for integrated circuit (IC) design are frequently the second largest budget cost for fabless IC design companies (after salaries)” (Byrne, 2007, p. 1). Traditional workload managers for technical computing optimize the utilization of hardware to ensure that hardware is kept busy with the most important job requiring hardware resources. But what about other resources? When we factor in the license cost, it becomes of critical importance to ensure that the software licenses are optimally used; including running on the most appropriate hardware and for the correct user/project. Therefore, the overall management of these costly assets is an important business consideration.
But how can you optimize use of these important assets?
The latest release of IBM Platform LSF, version 9.1.2, includes license scheduling as a standard feature which enables the monitoring of licenses managed by FlexNet. Platform LSF will throttle workload of a single Platform LSF cluster to avoid over subscribing a license. Furthermore it can also be used to view the license use of individual jobs.
IBM Platform License Scheduler - Standard Edition, which is available as an add on to Platform LSF, supports sharing of licenses amongst multiple Platform LSF clusters, as well as dynamically balancing license use among multiple clusters, projects. This ensures that license availability is prioritized by workload, user and project, and that licenses are optimally utilized. This leads to improved productivity, and the containment of costs.
To learn more about maximizing the agility of your distributed computing environment, check out the series of 20-minute webcasts here.
Byrne, G. (2007). EDA license management evolution. Retrieved from Edalics Web site: