The process of establishing a connection from one host to another is quite complex and involves multiple packet exchanges between two endpoints, which can be quite time consuming. The overhead of connection handshaking can be significant, especially for small HTTP messages. One can achieve a much higher data throughput if open connections can be re-used to execute multiple requests.

HTTP/1.1 states that HTTP connections can be re-used for multiple requests per default. HTTP/1.0 compliant endpoints can also use a mechanism to explicitly communicate their preference to keep connection alive and use it for multiple requests. HTTP agents can also keep idle connections alive for a certain period time in case a connection to the same target host is needed for subsequent requests. The ability to keep connections alive is usually refered to as connection persistence. HttpClient fully supports connection persistence.

HttpClient is capable of establishing connections to the target host either directly or via a route that may involve multiple intermediate connections - also referred to as hops. HttpClient differentiates connections of a route into plain, tunneled and layered. The use of multiple intermediate proxies to tunnel connections to the target host is referred to as proxy chaining.

Plain routes are established by connecting to the target or the first and only proxy. Tunnelled routes are established by connecting to the first and tunnelling through a chain of proxies to the target. Routes without a proxy cannot be tunnelled. Layered routes are established by layering a protocol over an existing connection. Protocols can only be layered over a tunnel to the target, or over a direct connection without proxies.

When equipped with a pooling connection manager such as PoolingClientConnectionManager, HttpClient can be used to execute multiple requests simultaneously using multiple threads of execution.

The PoolingClientConnectionManager will allocate connections based on its configuration. If all connections for a given route have already been leased, a request for a connection will block until a connection is released back to the pool. One can ensure the connection manager does not block indefinitely in the connection request operation by setting 'http.conn-manager.timeout' to a positive value. If the connection request cannot be serviced within the given time period ConnectionPoolTimeoutException will be thrown.

PoolingHttpClientConnectionManager cm = new PoolingHttpClientConnectionManager();
CloseableHttpClient httpClient = HttpClients.custom()
        .setConnectionManager(cm)
        .build();

// URIs to perform GETs on
String[] urisToGet = {
    "http://www.domain1.com/",
    "http://www.domain2.com/",
    "http://www.domain3.com/",
    "http://www.domain4.com/"
};

// create a thread for each URI
GetThread[] threads = new GetThread[urisToGet.length];
for (int i = 0; i < threads.length; i++) {
    HttpGet httpget = new HttpGet(urisToGet[i]);
    threads[i] = new GetThread(httpClient, httpget);
}

// start the threads
for (int j = 0; j < threads.length; j++) {
    threads[j].start();
}

// join the threads
for (int j = 0; j < threads.length; j++) {
    threads[j].join();
}

While HttpClient instances are thread safe and can be shared between multiple threads of execution, it is highly recommended that each thread maintains its own dedicated instance of HttpContext .

static class GetThread extends Thread {

    private final CloseableHttpClient httpClient;
    private final HttpContext context;
    private final HttpGet httpget;

    public GetThread(CloseableHttpClient httpClient, HttpGet httpget) {
        this.httpClient = httpClient;
        this.context = HttpClientContext.create();
        this.httpget = httpget;
    }

    @Override
    public void run() {
        try {
            CloseableHttpResponse response = httpClient.execute(
                    httpget, context);
            try {
                HttpEntity entity = response.getEntity();
            } finally {
                response.close();
            }
        } catch (ClientProtocolException ex) {
            // Handle protocol errors
        } catch (IOException ex) {
            // Handle I/O errors
        }
    }

}

One of the major shortcomings of the classic blocking I/O model is that the network socket can react to I/O events only when blocked in an I/O operation. When a connection is released back to the manager, it can be kept alive however it is unable to monitor the status of the socket and react to any I/O events. If the connection gets closed on the server side, the client side connection is unable to detect the change in the connection state (and react appropriately by closing the socket on its end).

HttpClient tries to mitigate the problem by testing whether the connection is 'stale', that is no longer valid because it was closed on the server side, prior to using the connection for executing an HTTP request. The stale connection check is not 100% reliable and adds 10 to 30 ms overhead to each request execution. The only feasible solution that does not involve a one thread per socket model for idle connections is a dedicated monitor thread used to evict connections that are considered expired due to a long period of inactivity. The monitor thread can periodically call ClientConnectionManager#closeExpiredConnections() method to close all expired connections and evict closed connections from the pool. It can also optionally call ClientConnectionManager#closeIdleConnections() method to close all connections that have been idle over a given period of time.

public static class IdleConnectionMonitorThread extends Thread {
    
    private final HttpClientConnectionManager connMgr;
    private volatile boolean shutdown;
    
    public IdleConnectionMonitorThread(HttpClientConnectionManager connMgr) {
        super();
        this.connMgr = connMgr;
    }

    @Override
    public void run() {
        try {
            while (!shutdown) {
                synchronized (this) {
                    wait(5000);
                    // Close expired connections
                    connMgr.closeExpiredConnections();
                    // Optionally, close connections
                    // that have been idle longer than 30 sec
                    connMgr.closeIdleConnections(30, TimeUnit.SECONDS);
                }
            }
        } catch (InterruptedException ex) {
            // terminate
        }
    }
    
    public void shutdown() {
        shutdown = true;
        synchronized (this) {
            notifyAll();
        }
    }
    
}

The HTTP specification does not specify how long a persistent connection may be and should be kept alive. Some HTTP servers use a non-standard Keep-Alive header to communicate to the client the period of time in seconds they intend to keep the connection alive on the server side. HttpClient makes use of this information if available. If the Keep-Alive header is not present in the response, HttpClient assumes the connection can be kept alive indefinitely. However, many HTTP servers in general use are configured to drop persistent connections after a certain period of inactivity in order to conserve system resources, quite often without informing the client. In case the default strategy turns out to be too optimistic, one may want to provide a custom keep-alive strategy.

ConnectionKeepAliveStrategy myStrategy = new ConnectionKeepAliveStrategy() {

    public long getKeepAliveDuration(HttpResponse response, HttpContext context) {
        // Honor 'keep-alive' header
        HeaderElementIterator it = new BasicHeaderElementIterator(
                response.headerIterator(HTTP.CONN_KEEP_ALIVE));
        while (it.hasNext()) {
            HeaderElement he = it.nextElement();
            String param = he.getName();
            String value = he.getValue();
            if (value != null && param.equalsIgnoreCase("timeout")) {
                try {
                    return Long.parseLong(value) * 1000;
                } catch(NumberFormatException ignore) {
                }
            }
        }
        HttpHost target = (HttpHost) context.getAttribute(
                HttpClientContext.HTTP_TARGET_HOST);
        if ("www.naughty-server.com".equalsIgnoreCase(target.getHostName())) {
            // Keep alive for 5 seconds only
            return 5 * 1000;
        } else {
            // otherwise keep alive for 30 seconds
            return 30 * 1000;
        }
    }

};
CloseableHttpClient client = HttpClients.custom()
        .setKeepAliveStrategy(myStrategy)
        .build();

HTTP connections make use of a java.net.Socket object internally to handle transmission of data across the wire. However they rely on the ConnectionSocketFactory interface to create, initialize and connect sockets. This enables the users of HttpClient to provide application specific socket initialization code at runtime. PlainConnectionSocketFactory is the default factory for creating and initializing plain (unencrypted) sockets.

The process of creating a socket and that of connecting it to a host are decoupled, so that the socket could be closed while being blocked in the connect operation.

HttpClientContext clientContext = HttpClientContext.create();
PlainConnectionSocketFactory sf = PlainConnectionSocketFactory.getSocketFactory();
Socket socket = sf.createSocket(clientContext);
int timeout = 1000; //ms
HttpHost target = new HttpHost("localhost");
InetSocketAddress remoteAddress = new InetSocketAddress(
        InetAddress.getByAddress(new byte[] {127,0,0,1}), 80);
sf.connectSocket(timeout, socket, target, remoteAddress, null, clientContext);

ConnectionSocketFactory plainsf = <...>
LayeredConnectionSocketFactory sslsf = <...>
Registry<ConnectionSocketFactory> r = RegistryBuilder.<ConnectionSocketFactory>create()
        .register("http", plainsf)
        .register("https", sslsf)
        .build();

HttpClientConnectionManager cm = new PoolingHttpClientConnectionManager(r);
HttpClients.custom()
        .setConnectionManager(cm)
        .build();

HttpClientContext clientContext = HttpClientContext.create();
KeyStore myTrustStore = <...>
SSLContext sslContext = SSLContexts.custom()
        .useTLS()
        .loadTrustMaterial(myTrustStore)
        .build();
SSLConnectionSocketFactory sslsf = new SSLConnectionSocketFactory(sslContext);

SSLContext sslContext = SSLContexts.createSystemDefault();
SSLConnectionSocketFactory sslsf = new SSLConnectionSocketFactory(
        sslContext,
        SSLConnectionSocketFactory.STRICT_HOSTNAME_VERIFIER);

Even though HttpClient is aware of complex routing scemes and proxy chaining, it supports only simple direct or one hop proxy connections out of the box.

The simplest way to tell HttpClient to connect to the target host via a proxy is by setting the default proxy parameter:

HttpHost proxy = new HttpHost("someproxy", 8080);
DefaultProxyRoutePlanner routePlanner = new DefaultProxyRoutePlanner(proxy);
CloseableHttpClient httpclient = HttpClients.custom()
        .setRoutePlanner(routePlanner)
        .build();

One can also instruct HttpClient to use the standard JRE proxy selector to obtain proxy information:

SystemDefaultRoutePlanner routePlanner = new SystemDefaultRoutePlanner(
        ProxySelector.getDefault());
CloseableHttpClient httpclient = HttpClients.custom()
        .setRoutePlanner(routePlanner)
        .build();

Alternatively, one can provide a custom RoutePlanner implementation in order to have a complete control over the process of HTTP route computation:

HttpRoutePlanner routePlanner = new HttpRoutePlanner() {

    public HttpRoute determineRoute(
            HttpHost target,
            HttpRequest request,
            HttpContext context) throws HttpException {
        return new HttpRoute(target, null,  new HttpHost("someproxy", 8080),
                "https".equalsIgnoreCase(target.getSchemeName()));
    }

};
CloseableHttpClient httpclient = HttpClients.custom()
        .setRoutePlanner(routePlanner)
        .build();
    }
}