SCIENCE

Puppet Labs has announced that the National Institute for Computational Science (NICS) is using Puppet as its systems management solution. NICS, a joint project between the University of Tennessee Knoxville and the Oak Ridge National Laboratory, is home to one of the world's most powerful computing complexes. With Puppet, NICS can better ensure the performance and availability of various systems such as secure essential infrastructure servers and high-performance computing systems.

NICS manages many systems under their current deployment of Puppet, including Kraken, the first academic petaflop supercomputer and the eighth fastest in the world. Accessed by 2,000 active researchers, Kraken contains 16,512 compute sockets, 129 terabytes of memory, and 3.3 petabytes of raw disk space. The XT5 system delivers in excess of 700 million CPU hours per year to the TeraGrid (a cyber infrastructure that offers more than a petaflop of computing power, over 30 petabytes of archival space, and access to over a 100 discipline-specific databases for research use).

High-performance computing administrators with NICS are responsible for installing, configuring and maintaining cutting edge technologies for use in the research community. Puppet gives NICS administrators confidence that they can provide a high level of reliability and recover rapidly from outages. Before Puppet, it used to take administrators four to six hours to deploy a system because they had to set up and maintain each server individually. With Puppet's holistic approach to configuration, the system is ready to go in about an hour. Puppet gives the administrators centralized control so they can apply changes consistently to all necessary servers and ensure that each of them has the appropriate level of security. The extra time allows the administrators to focus on maintaining an efficient infrastructure and keeping abreast of new updates.

"Twelve months ago we had no standard for managing our infrastructure; Puppet is now the standard. Our machines don't go up until they're in Puppet, tested, and working," said Stephen McNally, HPC administrator with NICS. "Being able to control everything we do in one place and ensure that it's consistent across the infrastructure gives us peace of mind. I don't have to worry about server one having a different security profile than server two -- that ultimately reduces our workload and allows us to deliver a reliable and secure infrastructure."

"We are delighted that NICS is using our systems management platform to maintain some of the world's most powerful computers," said Luke Kanies, CEO of Puppet Labs. "Puppet simplifies and speeds up the management of IT infrastructures, making it the perfect tool to ensure the reliability, consistency and security of even the largest, most complex infrastructures."