Merge pull request #240 from containous/update-benchmarks

update benchmarks with haproxy and latest results

.gitignore

@@ -8,3 +8,4 @@ traefik.toml
 *.test
 vendor/
 static/
+glide.lock

README.md

@@ -16,13 +16,14 @@ It supports several backends ([Docker :whale:](https://www.docker.com/), [Mesos/
 
 ## Features
 
+- [It's fast](docs/index.md#benchmarks)
 - No dependency hell, single binary made with go
 - Simple json Rest API
 - Simple TOML file configuration
 - Multiple backends supported: Docker, Mesos/Marathon, Consul, Etcd, and more to come
 - Watchers for backends, can listen change in backends to apply a new configuration automatically
 - Hot-reloading of configuration. No need to restart the process
-- Graceful shutdown http connections during hot-reloads
+- Graceful shutdown http connections
 - Circuit breakers on backends
 - Round Robin, rebalancer load-balancers
 - Rest Metrics

docs/index.md

@@ -1062,128 +1062,71 @@ Note that Træfɪk *will not watch for key changes in the `/traefik_configuratio
 
 ## <a id="benchmarks"></a> Benchmarks
 
-Here are some early Benchmarks between Nginx and Træfɪk acting as simple load balancers between two servers.
+Here are some early Benchmarks between Nginx, HA-Proxy and Træfɪk acting as simple load balancers between two servers.
 
 - Nginx:
 
 ```sh
-$ docker run -d -e VIRTUAL_HOST=test1.localhost emilevauge/whoami
-$ docker run -d -e VIRTUAL_HOST=test1.localhost emilevauge/whoami
+$ docker run -d -e VIRTUAL_HOST=test.nginx.localhost emilevauge/whoami
+$ docker run -d -e VIRTUAL_HOST=test.nginx.localhost emilevauge/whoami
 $ docker run --log-driver=none -d -p 80:80 -v /var/run/docker.sock:/tmp/docker.sock:ro jwilder/nginx-proxy
-$ ab -n 20000 -c 20  -r http://test1.localhost/
-This is ApacheBench, Version 2.3 <$Revision: 1528965 $>
-Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
-Licensed to The Apache Software Foundation, http://www.apache.org/
+$ wrk -t12 -c400 -d60s -H "Host: test.nginx.localhost" --latency http://127.0.0.1:80
+Running 1m test @ http://127.0.0.1:80
+  12 threads and 400 connections
+  Thread Stats   Avg      Stdev     Max   +/- Stdev
+    Latency   162.61ms  203.34ms   1.72s    91.07%
+    Req/Sec   277.57    107.67   790.00     67.53%
+  Latency Distribution
+     50%  128.19ms
+     75%  218.22ms
+     90%  342.12ms
+     99%    1.08s 
+  197991 requests in 1.00m, 82.32MB read
+  Socket errors: connect 0, read 0, write 0, timeout 18
+Requests/sec:   3296.04
+Transfer/sec:      1.37MB
+```
 
-Benchmarking test1.localhost (be patient)
-Completed 2000 requests
-Completed 4000 requests
-Completed 6000 requests
-Completed 8000 requests
-Completed 10000 requests
-Completed 12000 requests
-Completed 14000 requests
-Completed 16000 requests
-Completed 18000 requests
-Completed 20000 requests
-Finished 20000 requests
+- HA-Proxy:
 
-
-Server Software:        nginx/1.9.2
-Server Hostname:        test1.localhost
-Server Port:            80
-
-Document Path:          /
-Document Length:        287 bytes
-
-Concurrency Level:      20
-Time taken for tests:   5.874 seconds
-Complete requests:      20000
-Failed requests:        0
-Total transferred:      8900000 bytes
-HTML transferred:       5740000 bytes
-Requests per second:    3404.97 [#/sec] (mean)
-Time per request:       5.874 [ms] (mean)
-Time per request:       0.294 [ms] (mean, across all concurrent requests)
-Transfer rate:          1479.70 [Kbytes/sec] received
-
-Connection Times (ms)
-              min  mean[+/-sd] median   max
-Connect:        0    0   0.1      0       2
-Processing:     0    6   2.4      6      35
-Waiting:        0    5   2.3      5      33
-Total:          0    6   2.4      6      36
-
-Percentage of the requests served within a certain time (ms)
-  50%      6
-  66%      6
-  75%      7
-  80%      7
-  90%      9
-  95%     10
-  98%     12
-  99%     13
- 100%     36 (longest request)
+```
+$ docker run -d --name web1 -e VIRTUAL_HOST=test.haproxy.localhost emilevauge/whoami
+$ docker run -d --name web2 -e VIRTUAL_HOST=test.haproxy.localhost emilevauge/whoami
+$ docker run -d -p 80:80 --link web1:web1 --link web2:web2 dockercloud/haproxy
+$ wrk -t12 -c400 -d60s -H "Host: test.haproxy.localhost" --latency http://127.0.0.1:80
+Running 1m test @ http://127.0.0.1:80
+  12 threads and 400 connections
+  Thread Stats   Avg      Stdev     Max   +/- Stdev
+    Latency   158.08ms  187.88ms   1.75s    89.61%
+    Req/Sec   281.33    120.47     0.98k    65.88%
+  Latency Distribution
+     50%  121.77ms
+     75%  227.10ms
+     90%  351.98ms
+     99%    1.01s 
+  200462 requests in 1.00m, 59.65MB read
+Requests/sec:   3337.66
+Transfer/sec:      0.99MB
 ```
 
 - Træfɪk:
 
 ```sh
-docker run -d -l traefik.backend=test1 -l traefik.frontend.rule=Host -l traefik.frontend.value=test1.docker.localhost emilevauge/whoami
-docker run -d -l traefik.backend=test1 -l traefik.frontend.rule=Host -l traefik.frontend.value=test1.docker.localhost emilevauge/whoami
-docker run -d -p 8080:8080 -p 80:80 -v $PWD/traefik.toml:/traefik.toml -v /var/run/docker.sock:/var/run/docker.sock containous/traefik
-$ ab -n 20000 -c 20  -r http://test1.docker.localhost/
-This is ApacheBench, Version 2.3 <$Revision: 1528965 $>
-Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
-Licensed to The Apache Software Foundation, http://www.apache.org/
-
-Benchmarking test1.docker.localhost (be patient)
-Completed 2000 requests
-Completed 4000 requests
-Completed 6000 requests
-Completed 8000 requests
-Completed 10000 requests
-Completed 12000 requests
-Completed 14000 requests
-Completed 16000 requests
-Completed 18000 requests
-Completed 20000 requests
-Finished 20000 requests
-
-
-Server Software:        .
-Server Hostname:        test1.docker.localhost
-Server Port:            80
-
-Document Path:          /
-Document Length:        312 bytes
-
-Concurrency Level:      20
-Time taken for tests:   6.545 seconds
-Complete requests:      20000
-Failed requests:        0
-Total transferred:      8600000 bytes
-HTML transferred:       6240000 bytes
-Requests per second:    3055.60 [#/sec] (mean)
-Time per request:       6.545 [ms] (mean)
-Time per request:       0.327 [ms] (mean, across all concurrent requests)
-Transfer rate:          1283.11 [Kbytes/sec] received
-
-Connection Times (ms)
-              min  mean[+/-sd] median   max
-Connect:        0    0   0.2      0       7
-Processing:     1    6   2.2      6      22
-Waiting:        1    6   2.1      6      21
-Total:          1    7   2.2      6      22
-
-Percentage of the requests served within a certain time (ms)
-  50%      6
-  66%      7
-  75%      8
-  80%      8
-  90%      9
-  95%     10
-  98%     11
-  99%     13
- 100%     22 (longest request)
-```
+$ docker run -d -l traefik.backend=test1 -l traefik.frontend.rule=Host -l traefik.frontend.value=test.traefik.localhost emilevauge/whoami
+$ docker run -d -l traefik.backend=test1 -l traefik.frontend.rule=Host -l traefik.frontend.value=test.traefik.docker.localhost emilevauge/whoami
+$ docker run -d -p 8080:8080 -p 80:80 -v $PWD/traefik.toml:/traefik.toml -v /var/run/docker.sock:/var/run/docker.sock containous/traefik
+$ wrk -t12 -c400 -d60s -H "Host: test.traefik.docker.localhost" --latency http://127.0.0.1:80
+Running 1m test @ http://127.0.0.1:80
+  12 threads and 400 connections
+  Thread Stats   Avg      Stdev     Max   +/- Stdev
+    Latency   132.93ms  121.89ms   1.20s    66.62%
+    Req/Sec   280.95    104.88   740.00     68.26%
+  Latency Distribution
+     50%  128.71ms
+     75%  214.15ms
+     90%  281.45ms
+     99%  498.44ms
+  200734 requests in 1.00m, 80.02MB read
+Requests/sec:   3340.13
+Transfer/sec:      1.33MB
+```

middlewares/handlerSwitcher.go

@@ -0,0 +1,40 @@
+package middlewares
+
+import (
+	"github.com/gorilla/mux"
+	"net/http"
+	"sync"
+)
+
+// HandlerSwitcher allows hot switching of http.ServeMux
+type HandlerSwitcher struct {
+	handler     *mux.Router
+	handlerLock *sync.Mutex
+}
+
+// NewHandlerSwitcher builds a new instance of HandlerSwitcher
+func NewHandlerSwitcher(newHandler *mux.Router) (hs *HandlerSwitcher) {
+	return &HandlerSwitcher{
+		handler:     newHandler,
+		handlerLock: &sync.Mutex{},
+	}
+}
+
+func (hs *HandlerSwitcher) ServeHTTP(rw http.ResponseWriter, r *http.Request) {
+	hs.handlerLock.Lock()
+	handlerBackup := hs.handler
+	hs.handlerLock.Unlock()
+	handlerBackup.ServeHTTP(rw, r)
+}
+
+// GetHandler returns the current http.ServeMux
+func (hs *HandlerSwitcher) GetHandler() (newHandler *mux.Router) {
+	return hs.handler
+}
+
+// UpdateHandler safely updates the current http.ServeMux with a new one
+func (hs *HandlerSwitcher) UpdateHandler(newHandler *mux.Router) {
+	hs.handlerLock.Lock()
+	hs.handler = newHandler
+	defer hs.handlerLock.Unlock()
+}

server.go

@@ -48,7 +48,7 @@ type Server struct {
 
 type serverEntryPoint struct {
 	httpServer *manners.GracefulServer
-	httpRouter *mux.Router
+	httpRouter *middlewares.HandlerSwitcher
 }
 
 // NewServer returns an initialized Server.
@@ -82,7 +82,7 @@ func (server *Server) Start() {
 // Stop stops the server
 func (server *Server) Stop() {
 	for _, serverEntryPoint := range server.serverEntryPoints {
-		serverEntryPoint.httpServer.Close()
+		serverEntryPoint.httpServer.BlockingClose()
 	}
 	server.stopChan <- true
 }
@@ -142,22 +142,23 @@ func (server *Server) listenConfigurations() {
 				server.serverLock.Lock()
 				for newServerEntryPointName, newServerEntryPoint := range newServerEntryPoints {
 					currentServerEntryPoint := server.serverEntryPoints[newServerEntryPointName]
-					server.currentConfigurations = newConfigurations
-					currentServerEntryPoint.httpRouter = newServerEntryPoint.httpRouter
-					oldServer := currentServerEntryPoint.httpServer
-					newsrv, err := server.prepareServer(currentServerEntryPoint.httpRouter, server.globalConfiguration.EntryPoints[newServerEntryPointName], oldServer, server.loggerMiddleware, metrics)
-					if err != nil {
-						log.Fatal("Error preparing server: ", err)
-					}
-					go server.startServer(newsrv, server.globalConfiguration)
-					currentServerEntryPoint.httpServer = newsrv
-					server.serverEntryPoints[newServerEntryPointName] = currentServerEntryPoint
-					time.Sleep(1 * time.Second)
-					if oldServer != nil {
-						log.Info("Stopping old server")
-						oldServer.Close()
+					if currentServerEntryPoint.httpServer == nil {
+						newsrv, err := server.prepareServer(newServerEntryPoint.httpRouter, server.globalConfiguration.EntryPoints[newServerEntryPointName], nil, server.loggerMiddleware, metrics)
+						if err != nil {
+							log.Fatal("Error preparing server: ", err)
+						}
+						go server.startServer(newsrv, server.globalConfiguration)
+						currentServerEntryPoint.httpServer = newsrv
+						currentServerEntryPoint.httpRouter = newServerEntryPoint.httpRouter
+						server.serverEntryPoints[newServerEntryPointName] = currentServerEntryPoint
+						log.Infof("Created new Handler: %p", newServerEntryPoint.httpRouter.GetHandler())
+					} else {
+						handlerSwitcher := currentServerEntryPoint.httpRouter
+						handlerSwitcher.UpdateHandler(newServerEntryPoint.httpRouter.GetHandler())
+						log.Infof("Created new Handler: %p", newServerEntryPoint.httpRouter.GetHandler())
 					}
 				}
+				server.currentConfigurations = newConfigurations
 				server.serverLock.Unlock()
 			} else {
 				log.Error("Error loading new configuration, aborted ", err)
@@ -264,7 +265,7 @@ func (server *Server) startServer(srv *manners.GracefulServer, globalConfigurati
 	log.Info("Server stopped")
 }
 
-func (server *Server) prepareServer(router *mux.Router, entryPoint *EntryPoint, oldServer *manners.GracefulServer, middlewares ...negroni.Handler) (*manners.GracefulServer, error) {
+func (server *Server) prepareServer(router http.Handler, entryPoint *EntryPoint, oldServer *manners.GracefulServer, middlewares ...negroni.Handler) (*manners.GracefulServer, error) {
 	log.Info("Preparing server")
 	// middlewares
 	var negroni = negroni.New()
@@ -303,7 +304,7 @@ func (server *Server) buildEntryPoints(globalConfiguration GlobalConfiguration)
 	for entryPointName := range globalConfiguration.EntryPoints {
 		router := server.buildDefaultHTTPRouter()
 		serverEntryPoints[entryPointName] = serverEntryPoint{
-			httpRouter: router,
+			httpRouter: middlewares.NewHandlerSwitcher(router),
 		}
 	}
 	return serverEntryPoints
@@ -332,7 +333,7 @@ func (server *Server) loadConfig(configurations configs, globalConfiguration Glo
 				if _, ok := serverEntryPoints[entryPointName]; !ok {
 					return nil, errors.New("Undefined entrypoint: " + entryPointName)
 				}
-				newRoute := serverEntryPoints[entryPointName].httpRouter.NewRoute().Name(frontendName)
+				newRoute := serverEntryPoints[entryPointName].httpRouter.GetHandler().NewRoute().Name(frontendName)
 				for routeName, route := range frontend.Routes {
 					log.Debugf("Creating route %s %s:%s", routeName, route.Rule, route.Value)
 					route, err := getRoute(newRoute, route.Rule, route.Value)
@@ -443,7 +444,7 @@ func (server *Server) loadEntryPointConfig(entryPointName string, entryPoint *En
 	if len(entryPoint.Redirect.EntryPoint) > 0 {
 		regex = "^(?:https?:\\/\\/)?([\\da-z\\.-]+)(?::\\d+)(.*)$"
 		if server.globalConfiguration.EntryPoints[entryPoint.Redirect.EntryPoint] == nil {
-			return nil, errors.New("Unkown entrypoint " + entryPoint.Redirect.EntryPoint)
+			return nil, errors.New("Unknown entrypoint " + entryPoint.Redirect.EntryPoint)
 		}
 		protocol := "http"
 		if server.globalConfiguration.EntryPoints[entryPoint.Redirect.EntryPoint].TLS != nil {