MLNX_OFED (OpenFabrics Enterprise Distribution)

High-performance server and storage connectivity software
for field-proven RDMA and Transport Offload hardware solutions

DOWNLOAD SECTION

Overview

Use of clustered commodity servers, in lieu of traditional supercomputers and mainframes, offers tremendous price/performance benefits and unparalleled flexibility in deployment and long-term maintenance. To enable distributed computing transparently, high-performance computing (HPC) applications require the highest bandwidth and lowest possible latency. In enterprise data center (EDC) applications, these requirements are compounded with the need to support a large interoperable ecosystem of networking, virtualization, storage, and other applications and interfaces.

BENEFITS

• Single software stack that operates across all available InfiniBand devices and configurations such as mem-free, DDR/SDR, PCI-X, and PCI Express modes
• Support for HPC applications for scientific research, oil and gas exploration, car crash tests, bench marking etc. E.g., Fluent, LS-Dyna
• Support for EDC applications such as Oracle 10g RAC, IBM DB2, SAP, and financial applications
• Support for traditional IP and Sockets based applications leveraging the benefits of RDMA
• Support for high performance block storage applications utilizing RDMA benefits
• Support for cluster file systems for high performance and virtualized file system access
• Support for cluster management tool
  

MLNX_OFED 1.3 Enhancements

• Tighter integration and ease of installation with Linux offerings from Oracle, Red Hat and Novell to simplify and accelerate the deployment of InfiniBand and iWARP solutions
• Quality of Service (QoS) management capabilities that enable unified LAN, IPC and SAN applications over the same wire
• New routing algorithms to help optimize network and system performance and scalability
• Improved upper layer protocols for superior IP and sockets-based application performance, including database application
• Enhancements that enable MPI and parallel-processing applications to scale efficiently to large numbers of nodes while reducing the memory footprint required on servers
• Enhanced diagnostic tools to improve availability