Remote Direct Memory Access (RDMA) is an interesting way to improve network connections and bandwith by providing a direct memory access from the memory of one system into that of another. Compared to the full TCP/IP stack, RDMA can be managed without involving either one’s operating system (OS) and this means saving host resources and speed-up the communication.
RDMA permits high-throughput and low-latencynetworking, but more important is becoming a common feature on some network card, and also supported by different OSes and hypervisors.
But it’s essential a different protocol and could be difficult to troubleshoot it or just monitor it.
For system administrators looking to get the precise information on RDMA connections performance in their IT infrastructure, StarWind introduces Performance Benchmarking Tool (StarWind rPerf) – free RDMA link benchmarking tool. The utility measures both latency and bandwidth for RDMA connections between different operating systems such as Windows and Linux. No more need to maintain dozens of tools to find out the RDMA link performance and latency. StarWind rPerf is a single tool that does all the job.
Ideally, users benefit from avoiding useless cost, time, and effort wastes to support multiple solutions and get required information from various points.
StarWind rPerf is a single tool to measure both latency and bandwidth between Windows and Linux
StarWind rPerf is compatible with the following operating systems:
- Microsoft Windows: Windows 7 or later versions, Windows Server 2012 and later versions.
For Windows-based systems, NIC must have Network Direct Provider v1 and lossless RDMA configured. Latest drivers from NIC manufacturer are recommended, as standard Windows drivers don’t have ND API support. - Linux: CentOS 7 and Ubuntu distributions.
For Linux-based systems, latest network drivers with RDMA and RoCE support are required.
Moreover, StarWind rPerf is an all-in-one tool that measures both latency and bandwidth for cross-platform RDMA connections and this make the tool very useful to provide a complete visibility of this kind of implementation.
For more information see also: