ACADEMIA
Moab Workload Manager Claims Title as World’s First Petaflops Scheduler
In breaking the petaflops barrier, Los Alamos National Laboratories relied on Moab Workload Manager and TORQUE Resource Manager from Cluster Resources when running the milestone LINPACK Benchmark on Roadrunner. 
“We’re proud to be an integral part of this world-class record-setting system and to help extend the leading edge of workload-management scalability,” David Jackson, CTO of Cluster Resources, commented. “Congratulations to Los Alamos and system provider IBM for reaching this historic landmark.” Moab and TORQUE are workload-management software that optimize many of the world’s fastest supercomputers, including Lawrence Livermore National Laboratory’s IBM Blue Gene, which achieved 70.7 teraflops in 2004, making it the fastest computer in the world, a recognition it has maintained until now. Moab software also manages 40% of the world’s top 20 fastest supercomputers, as identified by the Top500 list published in November 2007. “Moab has been selected by many of the world’s biggest systems because of its ability to meet the extreme scalability and unique hardware needs required by these larger systems,” Jackson stated. “Our products focus on improving the way these systems behave, how they are shared, and ensuring that every possible cycle is squeezed out of them. For a system worth more than a hundred million dollars, every percent of improvement is invaluable.” Moab is applied to not only Top500 clusters but also those with as few as four CPUs. Systems of all sizes can expect to achieve 90-99% utilization with Moab and maximize the work that can be accomplished. Moab does this by orchestrating the scheduling, monitoring, policy management, and reporting of resources and usage to provide a centralized view, help fine tune system behavior and overcome failures. “Moab has to understand the lower level processes and understand which machines are working and which machines are broken and then steer around any failures and place the workload request properly,” added Jackson.
“We’re proud to be an integral part of this world-class record-setting system and to help extend the leading edge of workload-management scalability,” David Jackson, CTO of Cluster Resources, commented. “Congratulations to Los Alamos and system provider IBM for reaching this historic landmark.” Moab and TORQUE are workload-management software that optimize many of the world’s fastest supercomputers, including Lawrence Livermore National Laboratory’s IBM Blue Gene, which achieved 70.7 teraflops in 2004, making it the fastest computer in the world, a recognition it has maintained until now. Moab software also manages 40% of the world’s top 20 fastest supercomputers, as identified by the Top500 list published in November 2007. “Moab has been selected by many of the world’s biggest systems because of its ability to meet the extreme scalability and unique hardware needs required by these larger systems,” Jackson stated. “Our products focus on improving the way these systems behave, how they are shared, and ensuring that every possible cycle is squeezed out of them. For a system worth more than a hundred million dollars, every percent of improvement is invaluable.” Moab is applied to not only Top500 clusters but also those with as few as four CPUs. Systems of all sizes can expect to achieve 90-99% utilization with Moab and maximize the work that can be accomplished. Moab does this by orchestrating the scheduling, monitoring, policy management, and reporting of resources and usage to provide a centralized view, help fine tune system behavior and overcome failures. “Moab has to understand the lower level processes and understand which machines are working and which machines are broken and then steer around any failures and place the workload request properly,” added Jackson.