United Devices (UD) announced that integration of its wide-ranging suite of HPC management components now gives the company an ability to satisfy the unique requirements of large HPC environments via a comprehensive management solution.
United Devices' HPC Management Suite marks the culmination of years of focus on delivering solutions that address the most common HPC capacity and management challenges. The suite is comprised of five Grid solutions that align with the industry's most common pain points and deployment preferences.
The suite components can be used individually or in unison to leverage, control and manage multi-vendor HPC environments, with security and scalability extending up to hundreds of thousands of devices.
“As the HPC Grid market has grown, UD has assembled a set of practical solutions that tackle the most common HPC capacity and management challenges. These solutions were designed in a way that embrace existing HPC infrastructure investments, minimize end user disruption and deliver near-term ROI across a range of industries,” said Ben Rouse, UD's chief executive officer.
“Significant investment in product development and external acquisition of key technology components were required to complete the core set of solutions,” he said. “Today's announcement trumpets significant integration between these solutions, along with the enhanced implementation methodologies that now provide the overarching management capabilities demanded by large HPC environments.”
Individual solutions that make up the HPC management suite include:
- HPC Analytics: A comprehensive, heterogeneous capacity and utilization planning and analytics solution focused on the entire HPC environment.
- Virtual Cluster: A solution for creating virtual dedicated cluster nodes to provide vast, cost-effective compute power at the departmental level.
- Meta-Scheduler: A workload management solution that provides single job submission across heterogeneous clusters and the entire HPC environment.
- PC Grid: A solution that harnesses the unused capacity of desktop devices to provide unparalleled power for compute intensive applications.
- Cluster Scheduler: A distributed resource manager (DRM) solution for job scheduling and resource management in a dedicated cluster environment.
Rouse said that demand for HPC management solutions has increased dramatically over the last 12 months as the underlying Grid technology risk has decreased due to a critical mass of production implementations across a range of industries. In addition, a 70 percent year-over-year growth in cluster demand has exacerbated the management challenge. In fact, he noted, the problem has become so significant that it is now getting the attention of senior business leadership, including heads of research and development, financial executives and even mainstream IT.
“Our HPC management suite provides customers a palette of products, solutions, and implementation alternatives that can be combined to address the unique challenges inherent in every large HPC environment,” said chief technology officer Jikku Venkat. “The solution elements were architected to integrate in a seamless manner. For example, UD's meta-scheduler can use data gleaned from HPC analytics in real time to automate provisioning according to predefined SLAs.”
Venkat added that UD's solutions and methodologies derive from years of responding to real-life customer use cases and preferences, resulting in an approach that accommodates the political realities and Grid adoption impediments faced by all HPC organizations.