Virtualization for High-Performance Computing

Short Description
The specific demands of high-performance computing (HPC) often mismatch the assumptions and algorithms provided by legacy operating systems (OS) for common workload mixes. While feature- and application-rich OSes allow for exible and low-cost hardware configurations, rapid development, and exible testing and debugging, the mismatch comes at the cost of | oftentimes significant | performance degradation for HPC applications.

Website: www.research.ibm.com | Filesize: 76kb
No of Page(s): 4

Content
…
The requirements of high-performance computing (HPC) on an operating system (OS) significantly differ from typical server and workstation workloads. HPC applications have historically pushed the limits of CPU performance and memory size to run ever-larger problem sizes. Space and time multiplexing | as provided by hardware virtualization | is of no importance to HPC users. In this paper we argue that virtualization provides other benefits like the specialization of HPC operating systems while preserving legacy compatibility.
With virtualization, multiple operating systems can safely coexist on one physical machine. The machine is multiplexed by a small privileged kernel, commonly referred to as a hypervisor or virtual machine monitor (VMM), which provides the illusion of one or more real machines. While dating back to the early 1970’s [5,14], virtual machines lately have had significant attention for server consolidation in data centers. Consolidation of oftentimes highly underutilized servers greatly reduces hardware and maintenance cost. Virtualized devices and live migration of running operating systems decouple software installations from the physical hardware configuration. Coexistence of different versions of operating systems avoids incompatibilities, reduces testing and upgrade costs, and eliminates issues with con icting software packages.
…

Get the file Download here