This week Platform Computing (makers of LSF and other software products for HPC) announced the launch of a new portal product, and upgraded Platform LSF 7 and HPC Workgroup Manager.
First, Application Center
Platform Application Center is available as an optional capability for both Platform LSF 7 and Platform HPC Workgroup Manager, making the configuration and use of HPC applications in a heterogeneous HPC cluster simple for both end users and administrators. It allows administrators to easily customize and publish HPC software as a service for end users through an easy-to-use, self-service portal. Application interfaces for popular HPC applications are included out-of-the-box, along with a highly customizable web-based GUI template for administrators to create additional application interfaces for new HPC applications. It also provides the ability to manage jobs and application data, as an alternative to using command line syntax.
“HPC as a service” is a phrase I heard a bunch at SC09. I’m still trying to figure out what it means. I’d like to get some time with Application Center to figure out what it has that we haven’t seen with a million other portals. Of course, being shrink-wrapped is an advantage in itself: most of the other solutions have been science experiments. And not particularly popular science experiments at that: portals still aren’t in wide adoption outside of a few communities.
Platform LSF 7 Update 6 also adds more job scheduling functionality for large-scale and complex HPC environments, designed to make users more productive and cluster administration more intuitive. A new scheduling feature allows users to define a multi-phased resource reservation, allowing resource scheduling changes across a job’s entire lifecycle rather than basing resource reservations on peak-demand requirements.
…Platform HPC Workgroup Manager 1.1 is an advanced cluster management solution that combines all the cluster management tools necessary to easily deploy, run and manage HPC environments into one product set. These new capabilities fulfill current demand for improved application performance, workload management, provisioning and node management. The upgrade also includes the ability to dynamically switch operating systems on a compute node based on workload demand.