Computer networks are complex, involving multiple layers of hardware, software and protocols governing the transmission and reception of data. The network must simultaneously integrate many functions while maintaining compatibility of device connections. Network architecture is the name for the design principles that go into creating and building a computer network. Designing a network involves both logical and physical considerations.
Network architecture relies on different communication standards, or protocols, to connect the various devices on a network. Some of the most popular technologies include Ethernet, Fast Ethernet, token ring and Fiber Distributed Data Interface (FDDI). These technologies have made network interfacing and design more sophisticated. Using these technologies allows network architecture to build robust frameworks that transcend the limitations of old paradigms.
Ethernet consists of physical cabling and protocol standards that govern the transmission of data through the cabling. Basic Ethernet design involves all devices on a network communicating over a single cable. A local area technology, Ethernet works best connecting devices over very short distances.
A computer sending data along an Ethernet cable has to address it so the right device intercepts it. Every device reads a packet sent along an Ethernet channel, but only the device matching the address can use it. Ethernet has built-in collision detection, which protects the network from colliding packets. Additionally, Ethernet devices “listen” to the cable to make sure no other device is transmitting. It is like a reconnaissance team giving the all clear to proceed.
The Ethernet design first created by Robert Meltcalfe at Xerox Corporation in 1973 could only support up to 10 megabits per second (Mbps). Fast Ethernet can transmit data at the rate of 100 Mbps. Fast Ethernet started to be deployed on a large scale in the 1990s when higher local area network (LAN) performance was needed by businesses. A major advantage of Fast Ethernet is the ability to coexist and integrate with existing installations. The result has been for most networks to include both Ethernet and Fast Ethernet. Using so-called “10/100” adapters is what makes this possible. The adapter detects whether the cable is Ethernet or Fast Ethernet and adjusts accordingly.
Developed by International Business Machines, token ring is a widely used alternative to Ethernet and Fast Ethernet. In contrast to the wait-and-see approach to data transmission used in Ethernet, token ring utilizes an orderly method of strict rules. Data is transmitted through a ring of computers, from device to device until it reaches its intended destination.
The transmission process occurs through a token, which is special permission given to a device to transmit. The token races around the ring of computers until it finds one that wants to transmit. The transmitting computer gets rid of the token, replacing it with a data frame which completes its own journey around the ring, returning to the transmitting computer. Finally, that computer removes the frame and creates a new token, which re-starts the process.
FDDI is an optical communications standard, capable of transmitting data at the rate of 100 Mbps at a distance of up to 200 kilometers (around 127 miles). Network architectures that use FDDI serve thousands of end-users, linking vast areas of smaller networks and transmitting huge amounts of data. FDDI sends digital data over fiber-optic cables, which are capable of transmitting enormous data loads at high speed. They are analogous to the spine that serves as the conduit for the human body’s nervous system.
This article provides an overview of the following network protocols and technologies: PPP, TCP/IP, IPX/SPX, netBEUI, FTP, HTTP, NFS, SMTP, ICMP, Telnet and SSH.
Network design has advanced as computer technology has expanded and grown more powerful. Networks allow linked computers to share data. This basic functionality belies a complex interaction of a multiplicity of different rules, hardware, programs and connection standards. The most foundational aspects of networks are protocols. Network protocols define the conventions for communication between network devices such as a computer and a router. Multiple protocols exist, and each one came about because of improvements in basic design or as innovations in response to intractable problems.
Point-to-Point Protocol (PPP)
This is more properly described as a protocol suite than as a single protocol. Though considered to be part of Transfer Control Protocol/Internet Protocol (TCP/IP), PPP can be thought of as a protocol suite within the suite of TCP/IP. PPP facilitates IP transmission between network devices sharing a direct link. Through a series of link phases, a secure connection is established over which data may be transmitted. PPP actually works as an intermediate-level packet structure. It enables the transmission of higher-level protocols, such as TCP/IP, across a diversity of link environments. PPP is most commonly used for dial-up networking between modems.
Transfer Control Protocol and Internet Protocol are actually two separate protocols. They are described in one phrase because they are widely used together. TCP/IP can be better understood by referring to the Open Systems Interconnection Model (OSI Model). TCP represents the third layer, Network, while IP represents the fourth layer, Transport. Thus, TCP/IP refers to situations were TCP is used to transmit information across IP networks. IP moves data from node to node, and TCP moves data from client to server. TCP also provides redundancy in that it detects errors or lost data. TCP will retransmit data until a successful transmission is completed.
Internetwork Packet Exchange/Sequenced Packet Exchange (IPX/SPX) is the legacy networking protocol for Novell NetWare. However, it has evolved and is now used in Microsoft NT, Lotus Notes, Vista and Microsoft XP environments. IPX/SPX was eventually replaced with the more popular TCP/IP for NetWare, although IPX/SPX enjoys broad usage in other network environments. IPX is easy to configure and works well with small networks. It enables datagram services in packet-switched internetworks. Similar to IP, it has several key differences. SPX facilitates transport-layer services in conjunction with IPX’s routing capabilities. IPX addresses are hexadecimal numbers consisting of a network address and a hardwired node address.
Network BIOS Extended User Interface (netBEUI) was developed from Microsoft’s Network Basic Input Output System (netBIOS). netBIOS, in turn, was created based on Disk Operating System (DOS) BIOS, which extended the basic system with local area network (LAN) capabilities. netBEUI extends the networking features of netBIOS by formalizing the frame format for arranging information in data transmission. netBEUI works very well when limited to a single LAN. To connect to an external network with TCP/IP, the server providing the connection must be equipped with both TCP/IP and netBEUI. The server can then be programmed to choose between TCP/IP for external connections and netBEUI for internal LAN transmission.
File Transfer Protocol (FTP) is the single best protocol for transmitting files over the Internet. It sets the standard for other protocols. FTP uses TCP/IP to transmit data, making it compatible with virtually every network system in existence. FTP downloads files from servers and uploads files to servers from personal computers. FTP is commonly used for compuer to computer file transmission. Since it uses TCP/IP, anyone can download FTP and start transmitting files if they know the network address of the destination.
HyperText Transfer Protocol (HTTP) is the most widely used protocol to transfer audio, visual, graphics, text and other multimedia files over the Internet. HTTP runs over TCP/IP. Whenever a user starts a Web browser, it automatically uses HTTP to pull information from a server with TCP/IP. Hypertext is any text displayed on a webpage with a blue underlined link to another page. A user clicks on a hypertext link and the HTTP daemon, which is a part of all servers, detects and redirects the user to the new link. HTTP makes the Internet possible; TCP/IP merely fleshes it out.
Network File System (NFS) is a Unix protocol developed by Sun Microsystems. It treats remote files as if they were local files on a hard disk, allowing users to manipulate them. Importantly, users can access files across different types of computers. Using Virtual File System (VFS), which lies on top of TCP/IP, NFS lets computers act as clients when accessing remote files and also lets computers act as servers when another user is accessing their files. NFS obviously operates within a client/server framework. The advantage of NFS is letting users treat remote files as local files, which makes file sharing much easier.
Simple Mail Transfer Protocol (SMTP) is another ubiquitous Internet protocol like HTTP. SMTP facilitates the transfer of electronic mail or e-mail from computer to computer across the Internet. SMTP is used to send and receive messages between a mail client, usually an email application on the user’s computer and a mail server. SMTP interacts with Post Office Protocol (POP) or Internet Mail Access Protocol (IMAP) to retrieve e-mail from a mail server. Users need to configure both the POP or IMAP server and the SMTP server when installing and setting up an email application on their computer.
Internet Control Message Protocol (ICMP) is one of the major protocols of the IP suite. ICMP is an error detection and diagnostic protocol, essential for any IP set-up. ICMP can support packets containing error messages, which alert users to problems with network connections. A ping command, for instance, uses ICMP to test an IP address’s availability. ICMP messages usually report problems with the integrity of datagrams. To prevent infinite regression, no ICMP messages are ever sent about ICMP messages. ICMP is architecturally layered on top of IP. Error control is critical when dealing with the Network OSI layer.
Address Resolution Protocol (ARP) is rarely noticed by network administrators due to its simplicity. It operates at the second layer of the OSI model, translating IP addresses into physical hardware addresses. It does this to map an IP address to a link layer address whenever it is demanded. ARP links the IP and Ethernet network layers to facilitate data transmission. It is one of the most critical protocols for a network to use, and its simple operation and interface makes problems rare. ARP enables Ethernet adapters to transmit data between network devices by synchronizing different OSI layers to the same address.
Telnet and SSH
Telnet is a protocol that enables a user to connect to a remote computer. Most computer users have used telnet to connect to another computer. Command-line telnets exist on most computers; they are standard on nearly every Unix platform, Linux distro, and on Macs. Even Windows includes the program telnet.exe which can be accessed via the DOS command prompt interface. Another protocol that accomplishes the same task is Secure Shell (SSH). SSH uses much stronger cryptography than telnet to protect users against hackers and eavesdroppers. Telnet is older than SSH and uses minimal security to send and receive commands, making it more vulnerable. SSH is almost automatic, and telnet allows the user to pass some features onto the remote server itself.
Local area networks (LANs) are created to provide networking services to a group of computers that are relatively close to one another. A LAN can serve a single floor in an office building or a classroom in a school. The size of a given LAN can vary from one room to one building or even to a group of buildings, depending on the users’ needs. When using TCP/IP networking, LANs are most commonly set up as a single subnet with a clearly defined mask. LANs can be created based on wired or wireless connections. Today, wireless is most commonly used because of its convenience and ease.
What Is A WAN?
Wide area networks (WANs) span much larger geographical distances than LANs. The largest WAN of all is the Internet, which has a global reach. Multiple LANs often co-exist inside a WAN. Each subnet mask is connected to the WAN through a router or switch. A lot of devices can be used to create a WAN, from satellite dishes and radio towers to regular telephone lines. A WAN can use many different data link layer protocols, which hold the protocols for each LAN.
Internet Vs. Intranet
The Internet is a global “network of networks,” capable of connecting any computer on earth to any other computer. An intranet, by contrast, uses Internet Protocol technology to securely transmit data between computers located on a private network. Intranets use Ethernet, Wi-Fi, browsers and servers, but they are firewalled and not connected to any other network. Intranets are growing in popularity because they cost much less than proprietary protocols on private networks. An example of an intranet is an internal email network such as the ones used by businesses to allow employee communication.
What Is A VPN?
The need to reliably share information securely from multiple remote locations has given rise to virtual private networks (VPNs). A private network of computers is created through the Internet with “virtual” connections. The connections between nodes are virtual because they are secured from the Internet itself and merely pass through it. Thus, virtual private networks are completely secure even though they use the Internet. The VPN acts as a private tunnel through the Internet, safely transporting data between users. VPNs encapsulate data packets; that is, they place one data packet within another, ensuring the internal packet’s safe transport.
Peer-to-peer Vs. Client-server networks
The traditional Internet model is client-server. A client, usually a web browser, requests data from a server, which stores and hosts web pages. The server must authenticate each browser through certificates, even if no secure connection is being used. A peer-to-peer network is wholly different. Each computer stores and shares files and data equally. Peer-to-peer networks became well-known due to the phenomenon of online music piracy. Peer-to-peer networks cost next to nothing to create, but they have no security. The danger from viruses, worms and Trojan horses on a peer-to-peer network is infamous.
The OSI (Open Systems Interconnection) Model
The OSI (Open Systems Interconnection) model is a complex model that implements protocols in seven layers within a given network. Unlike the types of networks we’ve discussed thus far, OSI is not a type of network. However, now that we’ve covered a few basic types of networks, it is useful to begin putting the pieces together by examining the OSI model.
The model defines what path data has to travel through to get from one device to another in a network. The layers are divided into two sets: transport and application. The transport set consists of physical, data, network and transport. The application set contains session, presentation and application. The Physical layer defines the characteristics of network hardware such as connections and timing. In Data, a specific Physical protocol is assigned to data. Packet sequencing is also defined. Network figures out how the data will be sent to recipients. Finally, Transport is responsible for flow control, error checking and data recovery.
The Session layer handles all aspects of connecting to the device used by the data recipient. The Presentation layer takes information provided by the seventh and last layer, Application, and makes it intelligible to the lower layers. Lastly, Application interfaces with the operating system and thus the user whenever he chooses to carry out activities on the network.