By Klaus Gottschalk
The Leibniz Supercomputing Centre (LRZ), nestled on the outskirts of Munich in the town of Garching, was established 50 years ago by the Bavarian Academy of Science, to provide supercomputing resources to researchers and scientists across the Munich Scientific Network of universities.
Since then, the Centre has been the home of such systems as the HLRB and HLRB-II and has grown to become the premiere computing operations center for researchers across Europe, as they work to answer computational-intensive scientific questions.
In June 2012, the Centre flipped the switch on its fastest and most innovative supercomputer to date, the IBM-designed and built SuperMUC (the suffix, MUC, is derived from the Munich airport code). In addition to being the 6th fastest computer in the world, the 155,000-core system was built with an innovative cooling process that uses hot water to cool the thousands of microprocessors. By using hot water instead of compressed air and fans, typically used for cooling, the SuperMUC consumes 40 percent less energy than the previous supercomputer in use at the Centre – resulting in dramatic energy cost savings for LRZ.
Almost as impressive as the environmental milestones of the SuperMUC is the sheer processing power and ability to generate, manage and store massive amounts of information. I have worked on this project since its inception two years ago, and am still amazed at the enormous opportunities made possible by this system. Consider for a moment the vastness of its storage. The system, as it stands, includes 12 petabytes of permanent disk storage and 320 terabytes of main memory. For comparison, one petabyte is about the equivalent of all the content in the U.S. Library of Congress – times 100.
When people talk about Big Data, research centers like LRZ are ground zero. And when you deal with volumes of data that large, the first thing after establishing the permanent storage system is creating a backup and archiving plan to protect it against disaster. That’s exactly what IBM and our Business Partner, SVA System Vertrieb Alexander GmbH, tackled recently. We built a massive archiving and backup storage infrastructure to safely store the growing volumes of data being generated by the LRZ.
How massive? Leveraging IBM’s advanced and energy-efficient tape library systems, as well as several enterprise class disk systems equipped with SSDs, the backup infrastructure has a storage capacity of 16.5 petabytes, which is expandable to 40 petabytes.
To design such a powerful infrastructure that blends the most innovative supercomputing technology to date, with a technology such as magnetic tape, which is just as strategic today – if not more so – as it was 60 years ago when it revolutionized the computer industry, is inspiring. Then again, supercomputing truly does require super storage. And we built the template at the LRZ.