hurl provides a suite of network benchmarking and result analytic tools.

You can start using hurl with commands similar to Apache Bench ab, but as your needs grow, it exposes a framework for more complicated and large scale test plans, including distributing work across many machines.

go install github.com/pquerna/hurl/hurl

hurl http supports most of the basic arguments from Apache Bench. A simple example:

# hurl http -c 100 -n 10000 http://127.0.0.1/

Concurrency Level: 100
Time Taken: 867.486578ms
Document Length: 612

Percentage of the requests served within a certain time (ms)
 Min    2.318614ms
Mean    7.947689ms
 50%    6.629707ms
 66%    7.162523ms
 75%    7.548412ms
 80%    7.784365ms
 90%    8.542364ms
 95%    10.286918ms
 98%    13.771837ms
 99%    16.87096ms
100%    1.083432462s (longest request)

Would run 100 concurrent clients until 10,000 requests have been completed against http://127.0.0.1/.

hurl also can benchmark other services, like MongoDB:

hurl swift -u ${SWIFT_USER} -k ${SWIFT_APIKEY} load

• http
• etcd
• swift

Please open issues in Github for ideas, bugs, and general thoughts. Pull requests are of course preferred :)

hurl is licensed under the Apache License, Version 2.0