Network topologies for high performance computing: ethernet, InfiniBand and storage

ABSTRACT

The size of High Performance Compute Clusters has been growing rapidly, sometimes scaling to tens of thousands of nodes. A key aspect of HPC performance is the characteristics and performance of the network. Within HPC, each node within the cluster needs to be able to communicate with other resources - storage, for example - and with other nodes in the cluster for control and inter-process communications. As a cluster scales in size, careful consideration with respect to IPC, management and storage access is required to ensure the highest application performance and availability.This session will present best-practices for building large networks to support the most demanding HPC environments, and includes proven design recommendations for building scalable Ethernet, InfiniBand and Storage fabrics. Additionally, troubleshooting techniques, tips and tools to simplify management of infrastructure will be discussed in this session.

INDEX TERMS

Primary Classification:
  C. Computer Systems Organization
  C.2 COMPUTER-COMMUNICATION NETWORKS
      C.2.1 Network Architecture and Design
          Subjects: Network topology

Additional Classification:
  C. Computer Systems Organization
  C.2 COMPUTER-COMMUNICATION NETWORKS
      C.2.1 Network Architecture and Design

          Nouns: Ethernet
      C.2.5 Local and Wide-Area Networks
  C.4 PERFORMANCE OF SYSTEMS
      Subjects: Reliability, availability, and serviceability; Performance attributes

General Terms:
Design, Performance