Can you remember the first time you used the Internet? Probably not; and unlike 'morning glory', you didnt just wake up with it there. Logan Booker looks at the languages spoken by today's global networks, and why the're still using them.
ABOVE: A basic breakdown of the seven layers of the OSI architecture. To the left are the 'upper layers'; those closest and including the Application layer. Protocols on these layers are more interested in compiling information on the data to be sent, rather than insuring it gets over the medium. As you move 'down', or right, though, fragmentation begins and the protection of data becomes more relevant. This diagram also helps to show where the protocols in the TCP/IP suite lie, and what they are responsible for.
Like most basic computing technologies, the mystical stuff behind networks and the Internet are, by IT standards, dinosaurs. Lethargic, creaky beasts designed and programmed with only the past in mind. The only reason they still roam the Earth today is because of widespread and deeply-nested usage, and for the fact that they just work. We're afraid to replace them, and eventually, we will need to. How much longer can the World Wide Web exist until the meteor storm strikes and blacks out our online sun? More importantly, why have they lasted so long?
The idea for the 'Internet' can be traced back to the 1960s, when J C R Licklider was throwing around the idea of a 'Galactic network'. However, it wasn't until the US Department of Defense (DoD) decided it needed to link its research departments together that things took shape. A reliable and robust system was required that could transmit data in case communications equipment was damaged. TCP/IP was born -- after a small stint of NCP (Network Control Protocol), which couldn't handle the expanding workload, and ARPANET (Advanced Research Projects Agency), what was arguably the first 'Internet', came into being.
Proto-type
If you haven't guessed by the diagram opposite (or the intro), we're talking about network protocols, specifically TCP/IP, and on a lesser note, IPX/SPX. While many computer-savvy people will be familiar with these names, most won't understand how they fundamentally interact with network hardware and software.
TCP/IP and IPX/SPX are network protocols. While IPX/SPX is based on the International Standards Organisation's seven-layer (Application, Presentation, Session, Transport, Network, Data Link, and Physical) Open Systems Interconnection model for network software development, TCP/IP uses the DoD's four-layer (Application, Transport, Network, and Link) model, also known as the DARPA model. This four-layer model translates easily to the OSI model -- TCP/IP's Application layer covers OSI's Application, Presentation and Session layers; Transport and Network cover their respective layers, and Link takes care of the Data Link and Physical layers.
OSI is a loose (not officially required to follow) model designed so manufacturers could develop intercommunicative applications without having the concern of how they would talk to the layers above and below them, collectively referred to as the 'stack'. The point of OSI was to make networking transparent, depending on the application or protocol you were programming.
Stacks on
Before we get our hands filthy with protocols, it's good to know the differences between and the functions of the different OSI layers. From the top layer:
Application: The program-friendly layer that gets everything rolling. Protocols in the Application layer are responsible for handling raw data from user programs and handing it down to the lower layers. It also designates what protocols will be used to handle particular data in the lower layers.
Presentation: When data is passed to this layer it's prepared for presentation to the Application layer, so it can be understood by the higher-level protocols.
Session: This layer is responsible for creating and maintaining connections between systems. It also looks after connections between applications.
Transport: Perhaps one of the most important layers, as protocols here need to make up for the unreliability of lower layers (which consist of Ethernet and phone connections etc). The rules for the acknowledgement of data, confirmation of connections and packet sequencing are dealt with by protocols on this layer.
Network: It's between this layer and the Physical layer that things can go awry (lost and dropped packets). The Network layer is responsible for the organisation of data into packets and the transmitting of this data to other networks. The protocols on this layer control the transit of packets between hosts.
Data Link: This is an error-free link between the Network and Physical layers that organises data in preparation for the transport medium. This is where the last major encapsulation takes place.
Physical: The last layer of the ISO model, the protocols here break the packet data up so it can be transported, bit by bit, over the physical connection, be that twisted-pair, Copper wire or wireless. Once data is received by the target host, it works its way up from here to be reconstructed and finally presented on the Application layer.
As you can see, there will be many protocols at work to take data from an application and to turn it into 'protected' data that can be transported over a link. The process of 'wrapping' data up as it passes from layer to layer is called 'encapsulation'. For this reason, TCP/IP and IPX/SPX are not 'singular'; they consist of many different protocols that work on a variety of layers of the OSI model. For example, the TCP part of TCP/IP works on the Session and Transport layers, while the IPX part of IPX/SPX works on the Network layer.
Pre-packaging
Encapsulation is an integral part of the layer-layer relationship. It involves taking data from the layer above and applying a header (and in some cases, a trailer) to that data. Imagine a scrap of paper with a message written on it. Now, take that message and place it into an envelope with the address of the recipient as well as a return to sender address. Place this envelope into another envelope; however, this has a more complex address. And so on and so on. This is encapsulation. For TCP/IP, you could imagine the house number as the port, the street as the target address, and the city as the remote network.
While header (and trailer) contents vary from protocol to protocol, many contain specific information for their layer, as well as some sort of source and target address (an IP address for the Transport and Network layers, MAC address for the Data Link layer) and checksum or CRC (Cyclic Redundancy Check) to verify data. With some protocols, the checksum is used to simply check that the header isn't corrupt, but with others, such as TCP, the checksum is used to make sure the header and data is okay.
It isn't foolproof though -- roughly 1 in every 32,000 TCP checksums is incorrect. And no, there's no checksum for the checksum, unless you count error-checking by lower layer protocols.
TCP/IP
TCP/IP has its hands in every layer down to Physical (simply called 'Link' in the DARPA four-layer model). As mentioned before, TCP/IP is a suite of protocols - it has something for everything, from FTP (File Transfer Protocol), HTTP (Hyper Text Tran
Issue: 133 | February, 2012