performance

performance It brings together what traditionally has been siloed HPC environments," said Brian Connors, vice president of high performance computing at IBM. Typically, organizations with heavy engineering operations will have separate server farms used by individual groups of engineers, he said. That arrangement often leaves computing resources unused. For many businesses, pooling together multiple high-performance computing systems can have benefits beyond the improved efficiencies that come from using otherwise idle resources.

By combining all of a company's high-performance computing resources, individual engineering groups could have access to more cores and thus finish projects faster. "It's about getting the job done faster because you're throwing hardware at the job," Connors said. IBM is making the HPC Management Suite for Cloud software available after using it internally. More than 3,000 engineers around the world in its Systems Development organization shared resources as they developed the POWER7 processor family. It claims to have cut costs in half and reduced the design cycle by six months as a result of using the software. The product works on X86 machines as well as IBM's Power Platform running Linux operating systems.

It doesn't use virtualization, which engineers typically shun because it can slow down the work. IBM's workload management software levels the loads. Rebooting is not required as workloads shift, Connors said. IBM is also offering services for customers who might want help installing the suite and getting started using it. It will also offer industry-specific versions of the HPC cloud offerings starting with one that is optimized for electronics companies and automotive and aerospace manufacturers. IBM isn't the first to offer such software for better utilization of high-performance computing resources. In 2009, Platform Computing began offering tools to let its customers build private clouds out of multiple clusters.

performance