evio/benchmarks at reuseport - evio

History

Josh Baker 5554eb0d00 removed iris files from benchmarks		2017-11-06 05:22:35 -07:00
..
fasthttp-server	benchmarks	2017-11-02 18:31:36 -07:00
net-echo-server	benchmarks	2017-11-02 18:31:36 -07:00
net-http-server	benchmarks	2017-11-02 18:31:36 -07:00
out	removed iris from benchmarks	2017-11-06 04:48:32 -07:00
.gitignore	ec2 benchmarks	2017-11-03 17:14:09 -07:00
README.md	removed iris files from benchmarks	2017-11-06 05:22:35 -07:00
analyze.go	use bombardier instead of wrk	2017-11-04 06:26:00 -07:00
bench-echo.sh	benchmarks	2017-11-03 16:39:28 -07:00
bench-http.sh	removed iris from benchmarks	2017-11-06 04:48:32 -07:00
bench-redis.sh	benchmarks	2017-11-03 16:39:28 -07:00
bench.sh	benchmarks	2017-11-03 16:39:28 -07:00

README.md

evio benchmark tools

Required tools:

bombardier for HTTP
tcpkali for Echo
Redis for Redis

Required Go packages:

go get gonum.org/v1/plot/...
go get -u github.com/valyala/fasthttp
go get -u github.com/tidwall/redcon

And of course Go is required.

Run bench.sh for all benchmarks.

Notes

The current results were run on an Ec2 c4.xlarge instance.
The servers started in single-threaded mode (GOMAXPROC=1).
Network clients connected over Ipv4 localhost.

Like all benchmarks ever made in the history of whatever, YMMV. Please tweak and run in your environment and let me know if you see any glaring issues.