Networking Basics
Computers running on the Internet communicate to each other using either the Transmission Control Protocol (TCP) or the User Datagram Protocol (UDP), as this diagram illustrates:
When you write Java programs that communicate over the network, you are programming at the application layer. Typically, you don’t need to concern yourself with the TCP and UDP layers. Instead, you can use the classes in the java.net package. These classes provide system-independent network communication. However, to decide which Java classes your programs should use, you do need to understand how TCP and UDP differ.
TCP
When two applications want to communicate to each other reliably, they establish a connection and send data back and forth over that connection. This is analogous to making a telephone call. If you want to speak to Aunt Beatrice in Kentucky, a connection is established when you dial her phone number and she answers. You send data back and forth over the connection by speaking to one another over the phone lines. Like the phone company, TCP guarantees that data sent from one end of the connection actually gets to the other end and in the same order it was sent. Otherwise, an error is reported.
TCP provides a point-to-point channel for applications that require reliable communications. The Hypertext Transfer Protocol (HTTP), File Transfer Protocol (FTP), and Telnet are all examples of applications that require a reliable communication channel. The order in which the data is sent and received over the network is critical to the success of these applications. When HTTP is used to read from a URL, the data must be received in the order in which it was sent. Otherwise, you end up with a jumbled HTML file, a corrupt zip file, or some other invalid information.
TCP (Transmission Control Protocol) is a connection-based protocol that provides a reliable flow of data between two computers.
UDP
The UDP protocol provides for communication that is not guaranteed between two applications on the network. UDP is not connection-based like TCP. Rather, it sends independent packets of data, called datagrams, from one application to another. Sending datagrams is much like sending a letter through the postal service: The order of delivery is not important and is not guaranteed, and each message is independent of any other.
UDP (User Datagram Protocol) is a protocol that sends independent packets of data, called datagrams, from one computer to another with no guarantees about arrival. UDP is not connection-based like TCP.
For many applications, the guarantee of reliability is critical to the success of the transfer of information from one end of the connection to the other. However, other forms of communication don’t require such strict standards. In fact, they may be slowed down by the extra overhead or the reliable connection may invalidate the service altogether.
Consider, for example, a clock server that sends the current time to its client when requested to do so. If the client misses a packet, it doesn’t really make sense to resend it because the time will be incorrect when the client receives it on the second try. If the client makes two requests and receives packets from the server out of order, it doesn’t really matter because the client can figure out that the packets are out of order and make another request. The reliability of TCP is unnecessary in this instance because it causes performance degradation and may hinder the usefulness of the service.
Another example of a service that doesn’t need the guarantee of a reliable channel is the ping command. The purpose of the ping command is to test the communication between two programs over the network. In fact, ping needs to know about dropped or out-of-order packets to determine how good or bad the connection is. A reliable channel would invalidate this service altogether.
The UDP protocol provides for communication that is not guaranteed between two applications on the network. UDP is not connection-based like TCP. Rather, it sends independent packets of data from one application to another. Sending datagrams is much like sending a letter through the mail service: The order of delivery is not important and is not guaranteed, and each message is independent of any others.
Many firewalls and routers have been configured not to allow UDP packets. If you’re having trouble connecting to a service outside your firewall, or if clients are having trouble connecting to your service, ask your system administrator if UDP is permitted.
Understanding Ports
Generally speaking, a computer has a single physical connection to the network. All data destined for a particular computer arrives through that connection. However, the data may be intended for different applications running on the computer. So how does the computer know to which application to forward the data? Through the use of ports.
Data transmitted over the Internet is accompanied by addressing information that identifies the computer and the port for which it is destined. The computer is identified by its 32-bit IP address, which IP uses to deliver data to the right computer on the network. Ports are identified by a 16-bit number, which TCP and UDP use to deliver the data to the right application.
In connection-based communication such as TCP, a server application binds a socket to a specific port number. This has the effect of registering the server with the system to receive all data destined for that port. A client can then rendezvous with the server at the server’s port, as illustrated here:
The TCP and UDP protocols use ports to map incoming data to a particular process running on a computer.
In datagram-based communication such as UDP, the datagram packet contains the port number of its destination and UDP routes the packet to the appropriate application, as illustrated in this figure:
Port numbers range from 0 to 65,535 because ports are represented by 16-bit numbers. The port numbers ranging from 0 — 1023 are restricted; they are reserved for use by well-known services such as HTTP and FTP and other system services. These ports are called well-known ports. Your applications should not attempt to bind to them.
Networking Classes in the JDK
Through the classes in java.net , Java programs can use TCP or UDP to communicate over the Internet. The URL , URLConnection , Socket , and ServerSocket classes all use TCP to communicate over the network. The DatagramPacket , DatagramSocket , and MulticastSocket classes are for use with UDP.
Package java.net
The java.net package can be roughly divided in two sections:
- A Low Level API, which deals with the following abstractions:
- Addresses, which are networking identifiers, like IP addresses.
- Sockets, which are basic bidirectional data communication mechanisms.
- Interfaces, which describe network interfaces.
- URIs, which represent Universal Resource Identifiers.
- URLs, which represent Universal Resource Locators.
- Connections, which represents connections to the resource pointed to by URLs.
Addresses
Addresses are used throughout the java.net APIs as either host identifiers, or socket endpoint identifiers.
But, in most cases, there is no need to deal directly with the subclasses, as the InetAddress abstraction should cover most of the needed functionality.
About IPv6
Not all systems have support for the IPv6 protocol, and while the Java networking stack will attempt to detect it and use it transparently when available, it is also possible to disable its use with a system property. In the case where IPv6 is not available, or explicitly disabled, Inet6Address are not valid arguments for most networking operations any more. While methods like InetAddress.getByName(java.lang.String) are guaranteed not to return an Inet6Address when looking up host names, it is possible, by passing literals, to create such an object. In which case, most methods, when called with an Inet6Address will throw an Exception.
Sockets
Sockets are means to establish a communication link between machines over the network. The java.net package provides 4 kinds of Sockets:
- Socket is a TCP client API, and will typically be used to connect to a remote host.
- ServerSocket is a TCP server API, and will typically accept connections from client sockets.
- DatagramSocket is a UDP endpoint API and is used to send and receivedatagram packets.
- MulticastSocket is a subclass of DatagramSocket used when dealing with multicast groups.
Sending and receiving with TCP sockets is done through InputStreams and OutputStreams which can be obtained via the Socket.getInputStream() and Socket.getOutputStream() methods.
Interfaces
The NetworkInterface class provides APIs to browse and query all the networking interfaces (e.g. ethernet connection or PPP endpoint) of the local machine. It is through that class that you can check if any of the local interfaces is configured to support IPv6.
Note, all conforming implementations must support at least one NetworkInterface object, which must either be connected to a network, or be a «loopback» interface that can only communicate with entities on the same machine.
High level API
- URI is the class representing a Universal Resource Identifier, as specified in RFC 2396. As the name indicates, this is just an Identifier and doesn’t provide directly the means to access the resource.
- URL is the class representing a Universal Resource Locator, which is both an older concept for URIs and a means to access the resources.
- URLConnection is created from a URL and is the communication link used to access the resource pointed by the URL. This abstract class will delegate most of the work to the underlying protocol handlers like http or https.
- HttpURLConnection is a subclass of URLConnection and provides some additional functionalities specific to the HTTP protocol. This API has been superseded by the newer HTTP Client API.
The recommended usage is to use URI to identify resources, then convert it into a URL when it is time to access the resource. From that URL, you can either get the URLConnection for fine control, or get directly the InputStream.
URI uri = new URI("http://www.example.com/"); URL url = uri.toURL(); InputStream in = url.openStream();Protocol Handlers
As mentioned, URL and URLConnection rely on protocol handlers which must be present, otherwise an Exception is thrown. This is the major difference with URIs which only identify resources, and therefore don’t need to have access to the protocol handler. So, while it is possible to create an URI with any kind of protocol scheme (e.g. myproto://myhost.mydomain/resource/ ), a similar URL will try to instantiate the handler for the specified protocol; if it doesn’t exist an exception will be thrown.
By default the protocol handlers are loaded dynamically from the default location. It is, however, possible to deploy additional protocols handlers as services . Service providers of type URLStreamHandlerProvider are located at runtime, as specified in the URL constructor.
Additional Specification