Transfer 10Gbps http traffic over 1Gbps networks :)
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
cmd/httptp cmd/httptp: added letsencrypt support May 4, 2017
.gitignore Added travis.yml and .gitignore Oct 13, 2016
.travis.yml travis: skip go1.7 testing, since cmd/httptp depends on go1.8 May 4, 2017
LICENSE initial commit Sep 20, 2016
README.md Fixed benchmark results' formatting in README.md May 5, 2017
client.go rewrite on top of fastrpc Dec 9, 2016
client_test.go rewrite on top of fastrpc Dec 9, 2016
common.go rewrite on top of fastrpc Dec 9, 2016
server.go rewrite on top of fastrpc Dec 9, 2016
server_test.go rewrite on top of fastrpc Dec 9, 2016
server_timing_test.go Server: added ability to pipeline requests accepted from a single client Dec 5, 2016
ssl-cert-snakeoil.key
ssl-cert-snakeoil.pem Added support for encrypted connections Oct 11, 2016

README.md

Build Status GoDoc Go Report

httpteleport

Teleports 10Gbps http traffic over 1Gbps networks. Built on top of fastrpc.

Use cases

httpteleport may significantly reduce inter-server network bandwidth overhead and costs for the following cases:

  • RTB servers.
  • HTTP-based API servers (aka REST, JSON, JSON-RPC or HTTP-RPC services and microservices).
  • Reverse proxies.
  • Load balancers.

How does it work?

It just sends batched http requests and responses over a single compressed connection. This solves the following issues:

  • High network bandwidth usage
  • High network packets rate
  • A lot of open TCP connections

Unlike http pipelining, httpteleport responses may be sent out-of-order. This resolves head of line blocking issue.

Links

  • Docs

  • httptp - standalone single-binary reverse proxy and load balancer based on httpteleport. httptp source code may be used as an example of httpteleport usage.

FAQ

  • Q: Why httpteleport doesn't use HTTP/2.0?

    A: Because http/2.0 has many features, which aren't used by httpteleport. More features complicate the code, make it more error-prone and may slow it down.

  • Q: Why does httpteleport provide fasthttp- based API instead of standard net/http- based API?

    A: Because httpteleport is optimized for speed. So it have to use fasthttp for http-related stuff to be fast.

  • Q: Give me performance numbers.

    A: httpteleport achieves 200K qps on a single CPU core in end-to-end test, where a client sends requests to a local server and the server sends responses back to the client:

$ GOMAXPROCS=1 go test -bench=. -benchmem
goos: linux
goarch: amd64
pkg: github.com/valyala/httpteleport
BenchmarkEndToEndGetNoDelay1          	  300000	      4346 ns/op	  60.05 MB/s	       0 B/op	       0 allocs/op
BenchmarkEndToEndGetNoDelay10         	  300000	      4370 ns/op	  59.71 MB/s	       3 B/op	       0 allocs/op
BenchmarkEndToEndGetNoDelay100        	  300000	      4406 ns/op	  59.23 MB/s	       6 B/op	       0 allocs/op
BenchmarkEndToEndGetNoDelay1000       	  300000	      4457 ns/op	  58.55 MB/s	      24 B/op	       0 allocs/op
BenchmarkEndToEndGetNoDelay10K        	  300000	      5868 ns/op	  44.48 MB/s	     178 B/op	       1 allocs/op
BenchmarkEndToEndGetDelay1ms          	  300000	      4771 ns/op	  54.70 MB/s	      21 B/op	       0 allocs/op
BenchmarkEndToEndGetDelay2ms          	  200000	      7943 ns/op	  32.86 MB/s	      31 B/op	       0 allocs/op
BenchmarkEndToEndGetDelay4ms          	  200000	      7741 ns/op	  33.71 MB/s	      31 B/op	       0 allocs/op
BenchmarkEndToEndGetDelay8ms          	  200000	     10580 ns/op	  24.67 MB/s	      26 B/op	       0 allocs/op
BenchmarkEndToEndGetDelay16ms         	  100000	     16923 ns/op	  15.42 MB/s	      50 B/op	       0 allocs/op
BenchmarkEndToEndGetCompressNone      	  200000	      7899 ns/op	  33.04 MB/s	      31 B/op	       0 allocs/op
BenchmarkEndToEndGetCompressFlate     	  100000	     13257 ns/op	  19.69 MB/s	     129 B/op	       0 allocs/op
BenchmarkEndToEndGetCompressSnappy    	  200000	      8158 ns/op	  31.99 MB/s	      40 B/op	       0 allocs/op
BenchmarkEndToEndGetTLSCompressNone   	  200000	      8692 ns/op	  30.02 MB/s	      39 B/op	       0 allocs/op
BenchmarkEndToEndGetTLSCompressFlate  	  100000	     13710 ns/op	  19.04 MB/s	     131 B/op	       0 allocs/op
BenchmarkEndToEndGetTLSCompressSnappy 	  200000	      8480 ns/op	  30.78 MB/s	      42 B/op	       0 allocs/op
BenchmarkEndToEndGetPipeline1         	  300000	      4673 ns/op	  55.85 MB/s	       0 B/op	       0 allocs/op
BenchmarkEndToEndGetPipeline10        	  300000	      4610 ns/op	  56.61 MB/s	       3 B/op	       0 allocs/op
BenchmarkEndToEndGetPipeline100       	  300000	      4576 ns/op	  57.03 MB/s	       6 B/op	       0 allocs/op
BenchmarkEndToEndGetPipeline1000      	  300000	      4886 ns/op	  53.41 MB/s	      26 B/op	       0 allocs/op