Red Hat enhances trust and collaboration for the world’s most powerful supercomputer

Lawrence Livermore National Laboratory needed a high-performing computing environment to support the many national security requirements of its multidisciplinary teams and its unique role in national defense and security worldwide. The answer came in the form of the National Nuclear Security Administration’s (NNSA) first exascale supercomputer, El Capitan.

When complete, El Capitan will have a processing power of over 2 (double precision) exaflops per second and will be used by NNSA’s 3 principal laboratories: LLNL, Los Alamos National Laboratory, and Sandia National Laboratory. As well as LLNL’s main stockpile stewardship missions, it will also help to enable research in key areas, such as materials discovery, fusion science, and advanced manufacturing. Work performed on El Capitan and its unclassified “sister” system Tuolumne will also impact advances in clean energy, climate science, seismic modeling, and computational biology.

One of many factors that will contribute to El Capitan’s success is the choice of a main operating system. El Capitan needs to be capable of handling cloud-based applications, artificial intelligence, machine learning, internal systems, and next-generation infrastructures – all while meeting critical criteria for the protection of national security.

LLNL and Red Hat share a long history of collaboration. As a result, both understood the importance of aligning the security and mission-based needs of El Capitan while also making it accessible and sustainable in terms of the usability of its software.

LLNL also had extensive experience working with the capabilities of Red Hat Enterprise Linux, recognizing the foundational contribution of the enterprise platform to many of the world’s most powerful supercomputers which include several smaller existing LLNL systems.

By selecting Red Hat Enterprise Linux for El Capitan, LLNL and other labs could ensure that the new supercomputer would align with current and future systems and environments. It would also provide the operational flexibility and efficiency needed for day-to-day system management and maintenance.

“We chose Red Hat because of its release processes, its attention to detail, and commitment to security,” said Trent D’Hooge, Deputy Division Leader of Operations and System Software Lead for El Capitan. “Red Hat uses the same source code from end to end, which allows us to support our users and debug issues much more quickly.” 

By deploying Red Hat Enterprise Linux, LLNL had a trusted and highly versatile operating system (OS) that was capable of meeting the demands of El Capitan and Tuolumne and also provided an experience to which its employees were familiar. “Many people think you need a much more stripped-down OS than you would use in standard computers, but it’s easier for users and works just fine to use a standard operating system,” said D’Hooge. “All our supercomputers use Red Hat Enterprise Linux.”

“Red Hat’s processes mean patches can usually be applied to fix zero-day issues within a day, and that those same levels of security can also be applied to LLNL’s sister labs.

“Having a reliable OS means researchers can have confidence in the simulation results they’re running,” said Bronis R de Supinksi, CTO – Livermore Computing, LLNL. “Simulations of climate, earthquakes, new therapeutics, fusion energy – the list is practically endless.”

It’s a system that’s enabling further cooperation between NNSA’s labs and is also attracting interest and offers to share knowledge with external agencies.

“NASA reached out to us to find out if they could also base their HPC OS on the Red Hat Enterprise Linux that we use,” said D’Hooge. “We found that it was a good match, and now we collaborate a lot with NASA in addressing high-performance computing needs.”