High Performance Computing ( Hpc )

1768 Words8 Pages
Abstract High performance computing (HPC) means aggregation of computational power to increase the ability of processing large problems in science, engineering, and business. HPC on the cloud allows performing on demand HPC tasks by high performance clusters in cloud environment. Connection structure of nodes in HPC clusters should provide fast communication between nodes. Besides, it should provide scalability as well. A hypercube topology for connection nodes in a HPC cluster leads to have fast communications between nodes. In addition, a hypercube topology provides ability of scalability which is needed for high performance computing on the cloud. Introduction Before advent of internet, process and computing heavy workloads were done…show more content…
Large providers which have a vast number of resources are able to utilize their resources in a very efficient way. Furthermore public IaaS resources are utilized more efficient than private computer resources in average. Useful management approach for elastic virtual clusters, provide the capacity and environment according to specific requirements without the need to buy or use own hardware and software. While High Performance Cloud Computing model provides more flexibility, it faces several challenges: Sometimes companies have data and applications which consider as their property and they may not want to send it to shared resources. So some cloud providers offer special multi-tenant environments that is called virtual private clouds. Although it is hosted by the provider, customer has a full control on it. Some services offerings even go for shared-nothing resources to be exclusively used by a critical application [5]. However, the performance in High Performance Cloud Computing model is still an important issue: High Performance Computing applications needs more assurance of resource availability and timely result delivery. As in virtualize environment there is a higher I/O overhead and more request for using CPU cycles, it is hard to match the mentioned conditions. Besides, High Performance Computing jobs generally are CPU-bound, so they need thread synchronization during their executions. To achieve to a high performance, both the placement strategy and
Open Document