Future SOC Lab of Hasso Plattner Institute

Founded in June 2010, the Hasso Plattner Institute Future SOC Lab provides together with its international industry partners a high performance infrastructure for academic research, something usually only found at large enterprises.

"At HPI we have access to an experimental base that can rarely be financed in the university environment. This makes research possible that focuses on innovative concepts for IT systems of the future outside of the otherwise closed industrial laboratories," said Prof. Andreas Polze, head of the HPI research group OSs and middleware.

Scientists from universities as well as non-academic research institutes are invited to analyze highly complex IT systems and develop new ideas, data structures and algorithms in the HPI Future SOC Lab. The research may be followed up through the practical testing phase, according to Polze, who is also spokesman for the HPI Research School.

HPI’s modern high-performance systems have a 1,000-core cluster, an immense memory capacity in the terabyte range and large SSD-based storage systems.

HPI’s research lab provides, for example, the most current servers from Fujitsu and Hewlett-Packard, with four and eight Intel 64 bit CPUs and up to 128 arithmetic logic units, as well as 1TB to 2TB of main memory. Moreover, high-performance storage systems from EMC, smaller systems and a GPU computing system with NVIDIA Tesla and K20 units support a variety of research interests.

A unique opportunity on the academic landscape is provided with access to SAP’s new In-Memory appliance HANA, which is available for use in applications. SAP also offers the software-as-a-service enterprise solution ‘Business ByDesign’ with the development environment. Thanks to the HP Converged Cloud solution, HPI scientists have the possibility to establish rules concerning the allocation of computing resources for users of their top-level research lab and to automate operational processes.

Since the opening of the lab, over 100 researchers have completed more than 30 projects. The first research projects at the HPI have, for example, already led to significant advances in the acceleration of enterprise data analysis (In-Memory Data Management). Special procedures for service-oriented computing and reliable cloud computing are studied and developed in collaboration with industry partners for use in future systems.

Researchers at the Max Planck Institute for Molecular Plant Physiology in Potsdam are exploring together with HPI how metabolic networks of organisms can be examined faster through parallel simulation. With a powerful system based on graphic cards, HPI researchers investigate how 3-D scenes can be calculated in videos in real-time, in order to be dynamically adjusted to contents. Other projects deal with the management of VMs based on cloud computing to operate especially energy-efficient systems by switching off unused parts.