Benchmate

Overview

This repository contains

1. Network latency, throughput estimation tools
   • benchmate: iperf3, qperf like tool to estimate network throughput, latency.
   • konnectivity-benchmate: A client to measure throughput, latency of a Konnectivity setup.
2. Package to simplify construction of such tools. Read documentation.
3. HTTP handlers to quickly add network performance estimation to your services. See example.

Benchmate

benchmate measures latency and throughput between two nodes. You can run it on one node in the server mode and on another node in the client mode. If the the client and server can talk to each other, you should get the network performance stats at the client. This tool supports both TCP and Unix Domain sockets. Quickest way to try this out is using the docker image. If you are go programmer, you can build it yourself from cmd/benchmate directory.

Demo:

Commands from the demo:

# run server 
docker run --rm --network host quay.io/kubermatic-labs/benchmate

# run client
docker run --rm --network host quay.io/kubermatic-labs/benchmate -c

Konnectivity-benchmate

Client for benchmarking Konnectivity. You can run benchmark server on one node and point konnectivity-benchmate to the UDS of konnectivity proxy server.

Bmserver

This program demonstrates how you can easily add network performance estimation to your application. For example, if two microservices are communicating over a network, you can measure the latency and throughput of the network. You register HTTP handlers like pprof.

	mux := http.NewServeMux()
	mux.HandleFunc("/benchmate/throughput", bmHandler.Throughput)
	mux.HandleFunc("/benchmate/latency", bmHandler.Latency)
	log.Fatal(http.ListenAndServe(addr, mux))

You can then trigger handler on one service to run the server and on the other service to run the client.

Demo:

Commands from the demo:

# run the latency server 
curl http://localhost:8888/benchmate/latency --data '
{
    "msgSize": 128,
    "numMsg": 1000,
    "network": "tcp",
    "addr": ":13501",
    "timeout": 120000
}
'

# run the latency client 
curl http://localhost:9999/benchmate/latency --data '
{
    "msgSize": 128,
    "numMsg": 1000,
    "network": "tcp",
    "addr": ":13501",
    "timeout": 120000,
    "client": true
}
'

Troubleshooting

If you encounter issues file an issue or talk to us on the #kubermatic channel on the Kubermatic Slack.

Contributing

Thanks for taking the time to join our community and start contributing!

Feedback and discussion are available on the mailing list.

Before you start

• Please familiarize yourself with the Code of Conduct before contributing.
• See CONTRIBUTING.md for instructions on the developer certificate of origin that we require.
• Read how we're using ZenHub for project and roadmap planning

Pull requests

• We welcome pull requests. Feel free to dig through the issues and jump in.

Changelog

See the list of releases to find out about feature changes.

References

These programs are based on programs by Eli Bendersky (https://github.com/eliben). Original programs are in references folder.