

Qperf - Measure RDMA and IP performance

Quote

Qperf measures bandwidth and latency between two nodes. It can work over TCP/IP as well as the RDMA transports. On one of the nodes, qperf is typically run with no arguments designating it the server node. One may then run qperf on a client node to obtain measurements such as bandwidth, latency and cpu utilization.

In its most basic form, qperf is run on one node in server mode by invoking it with no arguments. On the other node, it is run with two arguments: the name of the server node followed by the name of the test.

With the qperf benchmark, you can measure the I/O performance of the network hardware and stack used in your Kubernetes cluster.

Mode of operation

As qperf requires a server and a client the controller creates the following objects during benchmark: * Server Deployment * Server Service * Client Pod

At the first step, the Server Deployment and Service are created. Once both becomes available, the Client Pod is created to execute the benchmark. Once the benchmark is completed (regardless of it's success), the server deployment and service is deleted from Kubernetes.

In order to avoid measuring loopback performance, it is advised that you set the affinity and anti-affinity scheduling primitives for the benchmark. The provided sample benchmark shows how to avoid executing the client and the server on the same machine. For further documentation please refer to Kubernetes' respective documentation.

Example configuration

You can find configuration example in the GitHub repository.

Sample benchmark

$ kubectl create --namespace kubestone -f https://raw.githubusercontent.com/xridge/kubestone/master/config/samples/perf_v1alpha1_qperf.yaml

Please refer to the quickstart guide for details on generic principles and setup of Kubestone.

Qperf Configuration

The complete documentation of qperf CR can be found in the API Docs.

Docker Image

Docker Image for qperf is provided via xridge's qperf-docker repository.

Legal

Qperf is licensed as GPLv2.