This week, Platform Computing launched Platform HPC, a cluster-management solution that’s geared towards small to mid-sized laboratories. In addition, the company said that it has enabled all of its products, including the latest release, to support graphic processing unit-based clusters and applications.
Platform HPC supports job submission templates for commonly used commercial applications and scripting guidelines for job submission as well as graphical processing unit job scheduling and expanded GPU monitoring capabilities.
Based on the same workload scheduler as Platform LSF, the company’s flagship product, Platform HPC also includes features such as workload management, reporting, monitoring, and message passing interface libraries.
However, while Platform LSF is designed to manage very large data centers with clusters of 64 nodes or more, Platform HPC aims to provide a solution for users with smaller clusters and less sophisticated data needs, Ken Hertzler, vice president of product management, told BioInform.
Although the tool isn’t geared primarily toward the life sciences, Hertzler expects that Platform HPC will find use among life science groups because many applications that are run in the space typically have a lot of processing requirements that are often performed in parallel on large clusters.
He noted that in the life sciences space many groups often build their own applications to solve their cluster-management needs because there aren’t many commercial options available and because some solutions are broken up into several separate components that need to be assembled, “which works great if you have an IT organization that is large and sophisticated enough to put all those pieces together,” he said
The company further claims that in addition to being a "single installer," Platform HPC is the "only complete solution” on the market and that provides an edge over other similar products offered by other vendors, which may have one or more features but lack others. For example, some vendors support only Windows-based clusters but not Linux-based ones, while others require users to install multiple components rather than offer a “one-step install.”
“Platform HPC is designed to focus on the market for an easier yet complete workload scheduling solution for smaller and medium-sized departmental users,” Hertzler said. “It has all the components you need in order to do cluster management … it also has all the functionality you need in a cluster … we provide applications to be integrated with the cluster and … we have made sure that all the components are included as a single installer and certified to work together.”
Platform HPC is available immediately and is marketed through hardware vendors such as Dell and Cray who provide clusters with the software preinstalled.
Licenses for Platform HPC are sold on a per-year basis and include support, maintenance, and upgrades to the software. Users can purchase either a workgroup edition or the enterprise edition.
The list price for the workgroup edition of Platform HPC, which has limited functionality and is restricted to a maximum of 32 nodes, is $250 per server. The enterprise model, which is intended for larger clusters and includes additional administrative capabilities and a floating license, is priced at $550 per server and available for an unlimited number of nodes.