Building the Right Environment to Support AI, Machine Learning and Deep Learning
Just as diplomats use diplomatic protocols in their meetings, computers use network protocols to communicate in computer networks. There are many network protocols in existence; TCP/IP is a family of network protocols that are used for the Internet.
A network protocol is a standard written down on a piece of paper (or, more precisely, with a text editor in a computer). The standards that are used for the Internet are called Requests For Comment (RFC). RFCs are numbered from 1 onwards. There are more than 4,500 RFCs today. Many of them have become out of date, so only a handful of the first thousand RFCs are still used today.
The International Standardization Office (ISO) has standardized a system of network protocols called as ISO OSI. Another organization that issues communication standards is the International Telecommunication Union (ITU) located in Geneva. The ITU was formerly known as the CCITT and, being founded in 1865, is one of the oldest worldwide organizations (for comparison, the Red Cross was founded in 1863). Some standards are also issued by the Institute of Electrical and Electronics Engineers (IEEE). RFC, standards released by RIPE (Réseaux IP Européens), and PKCS (Public Key Cryptography Standard) are freely available on the Internet and are easy to get hold of. Other organizations (ISO, ITU, and so on) do not provide their standards free of charge—you have to pay for them. If that presents a problem, then you have to spend some time doing some library research.
First of all, let's have a look at why network communication is divided into several protocols. The answer is simple although this is a very complex problem that reaches across many different professions. Most books concerning network protocols explain the problem using a metaphor of two foreigners (or philosophers, doctors, and so on) trying to communicate with each other. Each of the two can only communicate in his or her respective language. In order for them to be able to communicate with each other, they need a translator as shown in the following figure:
Figure 1.1: Three-layer communication architecture
The two foreigners exchange ideas, i.e., they communicate. But they only do so virtually. In reality, they are both handing over information to their interpreters, who then transmit this information by sending vibrations through the surrounding air with their vocal cords. Or if the parties are far away from each other, the interpreters communicate over the phone; thus the information is physically transmitted over phone lines. We can therefore talk about virtual communication in the horizontal direction (philosophical communication, the shared language between interpreters, and electronic signals transmitted via phone lines) and real communication in the vertical direction (foreigner-to-interpreter and interpreter-to-phone). We can thus distinguish three levels of communication:
- Between two foreigners
- Between interpreters
- Physical transmission of information using media (phone lines, sound waves, etc.)
Communication between the two foreigners and between the two interpreters is only virtual. In fact, the only real communication happens between the foreigner and his or her interpreter.
Even more layers are used in computer networks. The number of layers depends on which system of network protocols you choose to use. The system of network protocols is sometimes referred to as the network model. You most commonly work with a system that uses the Internet, which is also referred to as the TCP/IP family. In addition to TCP/IP, we will also come across the ISO OSI model that was standardized by the ISO.
Figure 1.2: Comparison of TCP/IP and ISO OSI network models
The TCP/IP family uses four layers while ISO OSI uses seven layers as shown in the figure above. The TCP/IP and ISO OSI systems differ from each other significantly, although they are very similar on the network and transport layers.
Except for some exceptions like SLIP or PPP, the TCP/IP family does not deal with the link and physical layers. Therefore, even on the Internet, we use the link and physical protocols of the ISO OSI model.
1.1 ISO OSI
Communication between two computers is shown in the following figure:
Figure 1.3: Seven-layer architecture of ISO OSI
1.1.1 Physical Layer
The physical layer is responsible for activating the physical circuit between the Data Terminal Equipment (DTE) and Data Circuit-terminating Equipment (DCE), communicating through it, and then deactivating it. Additionally, the physical layer is also responsible for the communication between DCEs (see Figure 1.3a). A computer or router can represent the DTE. The DCE, on the other hand, is usually represented by a modem or a multiplexer.
Figure 1.3a: DTE and DCE
To put it differently, the physical layer describes the electric or optical signals used for communicating between two computers. Physical circuits are created on the physical layer. Other appliances such as modems modulating a signal for a phone line are often put in the physical circuits created between two computers.
Physical layer protocols specify the following:
- Electrical signals (for example, +1V)
- Connector shapes (for example, V.35)
- Media type (twisted pair, coaxial cable, optical fiber, etc.)
- Modulation (for example, FM, PM, etc.)
- Coding (for example, RZ, NRZ, etc.)
- Synchronization (synchronous and asynchronous communication, time source, and so on)
1.1.2 Data Link Layer
As for serial links, the link layer provides data exchange between neighboring computers as well as data exchange between computers within a local network.
For the link layer, the basic unit of data transfer is the data link packet frame (see Figure 1.4). A data frame is composed of a header, payload, and trailer.
Figure 1.4: Data link packet or frame
A frame carries the destination link address, source link address, and other control information in the header. The trailer usually contains the checksum of the transported data. By using the checksum, we can find out whether the payload has been damaged during transfer. The network-layer packet is usually included in the payload.
In Figure 1.3a, the link layer does not engage in a conversation between DTE and DCE (the link layer does not see the DCE). It is engaged, however, in the frame exchange between DTEs. (It relies on the physical layer to handle the DCE issue.)
The following figure illustrates that different protocols can be used for each end of the connection on the physical layer. In our case, one of the ends uses the X.21 protocol while the other end uses the V.35 protocol. This rule is valid not only for serial links, but also for local networks. In local networks, you are more likely to encounter more complicated setups in which a switch that converts the link frames of one link protocol into link frames of a second one (for example, Ethernet into FDDI) is inserted between the two ends of the connection. This obviously results in different protocols being used on the physical layer.
Figure 1.5: Link layer communication
A serial port or an Ethernet card can serve as a link interface. A link interface has a link address that is unique within a particular Local Area Network (LAN).