Communication and Internet Technologies

📜 17.02 Protocols

The basic definition of a protocol is simple: it is a set of rules. Before asking what the rules are about, note three complications:

• We often say "a protocol" when we really mean a protocol suite that contains several individual protocols, because networking is complex enough to need many of them.
• A protocol may exist in several different versions.
• Some protocols are used to complement another protocol.

The rules have to be agreed by the sender and the receiver for any communication over a network. At the simplest level a protocol could say that a positive voltage represents a bit with value 1. It might set a maximum transmission speed. Many rules are about the format of a message or part of one, for example what the first 40 bytes of a packet must contain.

🥞 17.03 A protocol stack

The protocols in a suite can be viewed as layers in a protocol stack:

• Each layer can only accept input from the next higher layer or the next lower layer.
• A defined interface between adjacent layers is the only interaction allowed between them.
• A layer is serviced by the actions of the layers below it.
• Apart, possibly, from the lowest layer, each layer's functions are provided by installed software.
• A layer may be made up of sub-layers.
• Any user interaction uses protocols from the highest layer in the stack.
• Any direct access to hardware is confined to the lowest layer.

🌐 17.04 The TCP/IP protocol suite

TCP/IP is the protocol suite that underpins Internet use. It can be explained using the network model in Figure 17.02: TCP/IP only occupies the top three layers.

The diagram shows two end-systems and the logical connections between matching layers. An application running on one end-system behaves as though it had a direct connection to an application on the other. In reality:

1. The application-layer protocol on the sender sends a "message" to the transport-layer protocol on the same system.
2. The transport layer starts a process that delivers the identical "message" to the receiver end-system.
3. On the receiver, the transport layer hands the "message" up to the application-layer protocol.

The TCP/IP suite includes many protocols, for example:

The list shows how wide, and still evolving, the suite is. Only some of these protocols are studied further.

TCP (Transmission Control Protocol)

When an application on one end-system has a "message" for another end-system, an application-layer protocol passes the user data to the transport layer. TCP, working in the transport layer, is then responsible for the safe delivery of the message. It creates enough packets to hold all of the data; each packet is a header plus user data.

As well as safe delivery, TCP makes sure any response is directed back to the right application protocol. So the header carries a port number that identifies the application-layer protocol (for HTTP the port is 80), and the port number for the application at the receiving end-system. TCP is not concerned with the address of the receiving end-system. If the packet is one of a sequence, a sequence number is included so that the user data can be reassembled correctly.

TCP is connection-oriented. As in Section 17.01, only one packet of the sequence is sent to the network layer at first. When the network layer returns an acknowledgement that the connection has been established, TCP sends the other packets and receives response packets containing acknowledgements. That lets missing packets be identified and re-sent.

IP (Internet Protocol)

The job of the network layer, and of IP in particular, is to make sure of correct routing over the Internet. IP takes the packet from the transport layer and adds another header containing the IP addresses of both sender and receiver. To find the receiver's IP address it will very likely use the DNS service to look up the URL in the user data (Chapter 2, Section 2.09).

The IP packet, usually called a datagram, goes to the data-link layer, which belongs to a different protocol suite. The data-link layer puts datagrams into frames (see the next section). After that, IP has no further duty. IP is a connectionless service: once a packet has gone, IP does not know whether it arrived. If IP receives a packet containing an acknowledgement of an earlier packet, it simply passes it on to TCP without looking at the content.

The router

As Figure 17.02 shows, a frame sent by the data-link layer will arrive at a router during transmission, and more likely at several. There, the datagram inside the frame is handed back to IP. The router software then chooses the next target host, using a routing table for that router. The Internet is too big and complex for any router to hold a global routing table. Once the right address has been put into the datagram, IP passes it back down to the router's data-link layer.

Each router's routing table records any current problems with the possible next steps, so packets reach their destination in the shortest time available.

🔌 17.05 The Ethernet protocol stack

Chapter 2 (Section 2.05) introduced Ethernet as a protocol suite for a local area network (LAN). It can run in an isolated LAN with no outside connection, but nowadays a LAN is almost always connected to the Internet, so the LAN's protocol suite has to support the suite used for the Internet.

Look at the stack for one end-system in Figure 17.02. TCP/IP occupies the top three layers of the five-layer stack, so it is supported by the lower two layers. TCP/IP does not care how those two layers work: it is designed to run on whatever protocols are available.

Some sources show only a four-layer stack for an end-system. Either they leave out layers handled entirely by hardware rather than software, or they merge all of the support for TCP/IP into one layer.

Ethernet is the most likely protocol to provide the lower two layers. It can be seen as having two sub-layers for each of the Data link and Physical layers:

LLC (Logical Link Control), data link: handles interaction with the Network layer. It manages data transmissions and ensures their integrity. Ethernet is connectionless, though, so it is not responsible for checking that a transmission was delivered.

MAC (Medium Access Control), data link: assembles the Ethernet packet, called a frame. Two parts of the frame are the transmitter's address and the receiver's address. MAC also starts frame transmission and handles recovery from a failure caused by a collision (possibly using CSMA/CD).

PCS (Physical Coding Sublayer), physical: codes data ready for transmission, or decodes data received. It receives a frame from the MAC protocol or sends one to it.

PMA (Physical Medium Attachment), physical: transmits and receives the signals.

MAC addresses

Both addresses in the Ethernet frame are physical or MAC addresses. A MAC address uniquely identifies one network interface card (NIC), as in Chapter 2 (Section 2.04).

Addresses have stayed unique so far because the 48 bits used give enough different addresses. A 64-bit alternative is already used occasionally and is ready for when 48 bits run out. The 48-bit address is usually written in hexadecimal, for example:

4A:30:12:24:1A:10

In one version of the scheme, the first three bytes identify the manufacturer.

📨 17.06 Application-layer protocols associated with TCP/IP

There are very many application-layer protocols. These are some of the most used.

HTTP (HyperText Transfer Protocol)

HTTP underpins the World Wide Web, which makes it the most important application-layer protocol. Every time a user opens a website in a browser, HTTP is used, though you never see it working.

HTTP is a transaction-oriented, client–server protocol. The client sends a request message and the server sends back a response. HTTP defines the format of these messages. The first line of a request is the request line, which can be followed by header lines. All of it uses ASCII coding. The request line has the format:

<Method> <URL> <Version>CRLF

CR and LF are the ASCII carriage return and line feed characters. The method is usually GET, but there are several alternatives, which makes HTTP useful for much more than fetching web pages. The version must be given because HTTP has evolved and more than one version is in use.

Email protocols

The traditional way of sending and receiving email (Figure 17.04) involves three separate client–server interactions. The sender, acting as a client, connects to a mail server. That server then acts as a client when it sends the email on to the receiver's mail server. The receiver then collects it, acting as a client.

A more recent alternative to POP3 is IMAP (Internet Message Access Protocol):

• With POP3 emails are downloaded onto the client computer. You could argue this is safer from cyber-attack because the emails are stored locally, but a server is backed up regularly while a local client might not be. POP3 emails can only be read on that one client system.
• With IMAP the emails are not downloaded: they stay on the server but are accessible from the client. The big advantage is that the server can be reached from any client, which suits people on the move or using a different system from usual.

SMTP has largely been replaced by web-based mail. A browser is used to reach the email application, so the protocol used is HTTP. SMTP is still used to transfer email between the mail servers.

FTP (File Transfer Protocol)

For a routine file transfer from one user to another, the most likely method is attaching the file to an email. That is not always suitable. FTP is the application-layer protocol that can handle any file transfer between two end-systems. Transfer is not simple if the end-systems have different operating systems with different file systems. FTP handles this by separating the control process from the data-transfer process.

🧲 17.07 Peer-to-peer (P2P) file sharing

The traffic generated by peer-to-peer (P2P) file sharing is a major part of Internet use. P2P is an architecture with no structure and no controlling mechanism. Peers act as both clients and servers, and each peer is just one end-system. When a peer acts as a server it is called a seed.

The BitTorrent protocol is the most used, because it allows fast sharing of files. It has to solve three basic problems.

1. How does a peer find others that have the wanted content?
   Every content provider supplies a content description called a torrent. This is a file containing the name of the tracker (a server that leads peers to the content) and a list of the chunks that make up the content. The torrent file is at least three orders of magnitude smaller than the content, so it transfers quickly. The tracker keeps a list of all the other peers, called the swarm, that are actively downloading and uploading the content.
2. How do peers replicate content to give high-speed downloads for everyone?
   Peers download and upload chunks at the same time. They exchange lists of chunks and aim to download rare chunks first. Every time a rare chunk is downloaded, it automatically becomes less rare.
3. How do peers encourage others to provide content, rather than just download for themselves?
   This means dealing with free-riders or "leechers" who only download. A peer at first tries other peers at random, but then only keeps uploading to peers that provide regular downloads in return. A peer that does not upload, or uploads only slowly, will eventually be isolated or "choked".

BitTorrent has its own unusual vocabulary, and there are other terms not covered here. Fortunately, the principles are straightforward.

📌 Summary

• Circuit switching requires a dedicated circuit to be established between sender and receiver before transmission can start.
• In packet switching, packets can be transmitted without any circuit being created.
• Packet switching can use connectionless or connection-oriented transmission.
• A protocol suite is implemented as a layered stack.
• The TCP/IP protocol suite supports usage of the Internet.
• Ethernet is the most likely option for use in the Data link and Physical layers.
• Examples of application-layer protocols are HTTP, SMTP, POP3, IMAP and FTP.
• Peer-to-peer file sharing on the Internet uses the BitTorrent protocol.

🃏 Key-term flashcards

Tap a card to flip it.

✅ Quick quiz

📝 Exam-style questions

Write your answer first, then open the model answer. These are my mark-scheme-style points, not official mark schemes.

Application  ┐
Transport    ├ TCP/IP suite
Network      ┘
Data link    ┐  LLC sub-layer
             │  MAC sub-layer
Physical     │  PCS sub-layer
             ┘  PMA sub-layer   ← Ethernet suite
     ↕ transmission medium ↕