Networks and the Internet (Part 1)Article contents:
In the 1960s, the U.S. government was researching computer networks. This fell to ARPA, the Advanced Research Projects Agency. Part of the design requirements for this network was one that would be able to withstand disruptions in the physical infrastructure of the network.
The idea was that the military, using a communications network, could possibly have part of it destroyed during action. Early network designs were not “fault-tolerant”: a disruption in part of the network caused the entire network to crash, or stop working. Eventually, ARPA started using TCP/IP for its network, the ARPANET. This technology was fault-tolerant: TCP/IP could direct data around damaged parts of a network to reach the undamaged parts. This network eventually was opened up to universities to do research. Large educational institutions like Stanford and USC became part of the network. Not long after that, commercial entities saw the value of a large interconnected network (internet) and the general public was able to access this network through a university account or through internet service providers (ISPs) such as America Online (AOL).
The World Wide Web is often confused for the internet, yet the web is a set of services that are accessed through the internet. It requires a web browser to access these services. There are other data services on the internet that do not require a browser: online games such as World of Warcraft are good examples. It is interesting to note that internet speeds in the United States are slower than most of the rest of the world since the internet developed there first and existing copper cable and phone lines were used to send the data signals. Later, better technologies were developed such as fiber optic cable and countries in Europe and Asia installed those, while the U.S. is only slowly catching up. Since internet is already available in most locations, upgrading to fiber is seen as more of a luxury that a necessity in the U.S.
Wired and Wireless Networks and Topologies
All computer data are stored in binary, i.e., zeroes and ones. The data sent over networks is no exception. Electrical signals over a wire are created to symbolize either a one or a zero. By themselves, the numbers don’t mean very much, but in the proper context, the data is interpreted by the recipient computer.
Before there were wireless networks, all network data was sent over wires. There originally were different ways to design the layout of the network or the physical topology. Over the years, these have been simplified, and only one or two topologies are primarily used over local networks and the internet.
A LAN, or local area network, consists of all the computers within a specific physical location, such as an office building or floor or a home. Once you connect a LAN to another LAN over a large distance, it is called a WAN, or wide area network. You can imagine two offices from the same company, one in Madrid and one in London, connected together; this is an example of a WAN. The internet is the largest WAN, connecting millions of LANs together through wired and wireless connections.
Using a wired network naturally requires cabling. Standard wiring for a LAN (inside a home or an office) consists of UTP cable of various categories. Data speed for networks is usually measured in bits per second (bps.) Occasionally, you may see bytes per second (Bps); keep in mind this is 8 times more data throughput than bits per second.
The categories of UTP cable are as follows:
|Category 5||Up to 100Mbps (Megabits per second)|
|Category 5e||Up to 1000Mbps (1Gbps)|
|Category 6||Up to 10Gbps|
These UTP cables appear physically identical. You must usually look for a label or printing on the wire to see the category if you are not sure.
The type of connector used for UTP cabling is called an RJ-45 connector (shown above). Though it is similar to the standard U.S. telephone connector, it is actually larger and incompatible with phone wire, which is Category 1 and unusable for networks.
The simplest network connection device is a hub. The hub takes a signal from one cable and splits it out over several cables. If it was splitting out four signals, it would be called a “four port” hub. A hub does not do any signal processing—everything that comes in is sent out through all of the ports. Communication is two-way: the ports can also send communication through the uplink connection.
A switch is very similar to a hub. It has a connection for an uplink and multiple ports for communication. However, it does do basic filtering on data. Computers send data over a network in data frames with hardware addresses. Some data is sent out with the “broadcast” address; this address means it is for all computers who can receive it. A switch filters these out so they do not get propagated across the entire network (which can cause a flood of data known as a “broadcast storm”).
Switches are often placed together in a central location in an office building to split the signal out to different rooms and floors.
The router is the most intelligent of the three connecting devices. It examines the IP address of each packet of data sent over a network and only sends the signal through the connection that will reach the address. Since it filters data, it can also block data from entering a network that does not belong there; this function is called a firewall.
Wired Network Topologies
A network topology is the physical layout of the connecting devices used. In early days of networks, several different topologies were used. Today, a large portion of networks use the same design: the star topology. Some network topologies are shown and described below.
In this example, multiple star networks are connected through a larger star network. Switches for the center for each small network at the bottom of the figure. They connect five computers (or other devices) to the uplink, which goes up to another switch. Each switch at the second level has four connections, so it also forms its own star. The router sits at the edge of the private network, filtering data to and from the internet. The router is also the center of a star with two connections going to local switches and one going out to the internet. This is a common design within office buildings. Each second-level switch would handle the data from one floor of the building, where each lower-level switch is place in separate rooms with workers.
Wired Meets Wireless
Wireless networks were made available to the public in the 1990s. These networks help connect devices to a larger network. Inevitably, that network eventually connects to wires. Though a device you are using may be “wireless,” to reach devices at other geographical locations the signal will eventually have to be translated and sent over a wired network. Wireless networks have advantages and disadvantages compared to wired.
Obviously it is easier to carry your device around if it is untethered from a cable. However, if the signal is weak in wireless, it is often hard to find the cause and where the signal is strong in a room—it is impossible to tell just by looking. Throughout the history of wired and wireless networks, wired networks have always had the advantage in data speeds, though wireless is slowly catching up. A wired-only network can only be accessed by plugging into a cable, which allows for greater security, while anyone within range of the signal can connect (or attempt to connect) to Wi-Fi. Wireless is often easier to set up without the need to run cables.
The IEEE, or Institute of Electrical and Electronics Engineers, is a professional association of engineers. They have provided technical standardization for many aspects of electronics for decades. One of these standards is for wireless networks.
The original standard, set in 1997, is IEEE 802.11. New, updated standards were then introduced by adding additional letters after the name. There have been over a dozen standards in use, but there are three that are still in prominent use today.
The 802.11g protocol is the oldest widely used protocol; you can see that its theoretical speeds are over ten times slower than the next protocol available. However, just as it is stated, these speeds are theoretical—in practice, most networks do not achieve these maximum possible speeds. The 2.4GHz channel (frequency) for data can travel farther and through physical obstacles better than the 5GHz channels, but 5GHz frequencies can carry more data. 802.11n is able to switch between both frequencies. 802.11ac is the fastest current wireless protocol, but it is also the most expensive. It is backwards compatible with older devices, but they will have to operate at the slower protocol speed.
Keep in mind that for a network to use a particular protocol, all wireless devices must be able to connect. In other words, if you buy an 802.11ac Wi-Fi access point, all of your network devices must have network cards that support it. If not, you will have to buy new network cards to take advantage of the higher speeds.
The OSI Model and TCP/IP
To explain the process of network communication, OSI, the Open Systems Interconnection standard, was created. It is used to create specific networking technologies and provide a map of how those technologies work. The main description of this is the OSI Model. It is divided into seven layers.
Here’s a basic overview of how it works: the application layer communicates with applications that want to use the network (usually requested directly by a user). It takes the data and passes them to the presentation layer. Then the data are formatted into the correct encoding (for instance, Unicode). This is also where encryption is applied to the data. After that, a session is established with the other computer. The session layer establishes communication settings, such as ports, and synchronization of data. Once the settings are established, the transport layer handles the error checking and actual transmission of data, monitoring communications from end-to-end. The network layer is more specific; it handles the physical route of the packet each step along the way of the network. This is where IP addressing is mostly used. The data link layer exists to provide specifications for devices—such as Ethernet—that determine the way signals are sent electronically over the network, as well as standards for devices; this is where the hardware (MAC) address is used. The physical layer consists of the physical devices and wiring of the network, as well as the signal itself.
This layer is the closest to the user, yet it interacts with applications that request network access. Technically, this layer is for applications, not made of applications. It is the interface layer where applications make requests of the operating system to access the network. Once the request is accepted and delivered, it goes to the presentation layer. Applications that request network access can include internet apps such as browsers and other local apps using the network, such as MS Word. If you print a document to the network printer in the office, Word will request access through the application layer.
This layer prepares the data for being transferred. You might say that it being ready to be “presented.” Encoding happens here. For instance, if you are sending an email in Unicode, it will translate the symbols at this layer. Encryption occurs at this level. In other words, a plaintext or other unencrypted message is brought through a mathematical algorithm that created encryption. Without knowing the encryption key, the code is nearly unbreakable (though no code is perfect). The sending and receiving network devices share a key at the beginning of communication that allows for the algorithm to be reversed and the message to be turned back into the original.
Imagine you want to have a meeting with someone: you will set the time and place and some other requirements (e.g., bring your notebook). This is similar to a session for network communications. The start of communication is agreed upon, along with wait times for responses (acknowledgements) and how the session will be ended (terminated). Once communication is begun, lower layers handle the details.
Inside a session, the transport layer does the important work of managing the data transfer between hosts. It makes sure that the data flows at the proper rate and performs some error checking. It also breaks down the data into chunks, or segments, that are sent out separately.
If you wish to mail a package to a friend, there are several steps. You put it into a box (the previous layers do this) and give it an address. This address is what the network layer uses to transport the data (your box) to its destination. It does this by using routing. You may remember a device called a router—this is what performs this task. At each step along the way, the address is checked and the correct path is chosen to send the data further along toward its destination. In TCP/IP this is where the IP address is used. Data sent at this level is organized into packets.
Data Link Layer
Once the data makes it to its destination network, it still must find the exact device it is looking for. The data link layer works at a more specific level than network. It uses hardware addresses to identify network devices. This is the layer where Ethernet specifications dictate how signals are formed as well. It includes local network error checking, including collisions. Collisions occur when two network devices send out a signal simultaneously, and they collide over physical media. Most networks use collision detection. This means after a collision is detected both devices are instructed to wait a random amount of time before sending again.
Network devices are included in this layer such as hubs, switches, and routers as well as cabling. The signal itself is included at this layer, where zeros and ones in data are formed into a specific signal and placed on the network.
Data travels back and forth on a network from physical to application and back again. The OSI is a general model of how network data transmission works, but it is not a required list; different protocols use different concepts from the model.
Ethernet is a set of standards (defined by IEEE 802.3) used for LANs and WANs. The word “Ethernet” is used commonly to refer to parts of a local wired network, such as Ethernet cabling (Cat 5, 5e, and 6), and Ethernet ports, where the cables are connected. The standards of Ethernet include the physical layer, which gives specifications for devices and signals, and the data link layer, where it defines how signals are organized.
TCP/IP and the OSI model
TCP/IP stands for Transmission Control Protocol/Internet Protocol. The transmission control protocol works at the transport layer, while the internet protocol works at the network layer of the OSI model. However, TCP has its own four-layer model of operating.
TCP/IP is the protocol of the internet. That means that to access the internet, it must be installed on a device. Since it is also usable for LAN communication, most computers use TCP/IP as their primary and only network protocol. It consists of other sub-protocols that exist within it to facilitate communications. To use TCP/IP a network device must have three settings configured. These settings are 32-bit binary numbers (or 128-bit with the new IP version 6).
This is the unique address of the network device; there cannot be a duplicate on the same network. It is divided into two parts: a network ID (which identifies what network segment the device is on) and the host ID (which is the individual device number). A good analogy is that the network ID is like a street and the host ID is like a house number.
This is a series of ones and zeros that identify which part of the address is the network ID and which is the host ID. IP addresses have the option of being split in different places depending on the need of the network.
The gateway is the router or device that allows the computer to access other devices outside of its local network. If a computer is talking to a device on its own network it does not need a gateway, but once it wants to go outside of the local network it must use a gateway.
Let’s return to our analogy of sending a box in the mail, with more details. Let’s say you are sending a picture of your cat over Skype to your grandmother in Canada. The data is formatted and presented for transport over the internet. Then a session is established. The picture of your cat data is then broken down into individual packets. Each packet is like a separate box of data. The box will have two addresses on it: the sender’s and the recipient’s.
Granny’s IP address will be used to deliver the cat picture pieces to her. The packet goes to your router (the default gateway), which then sends it out through your internet service provider (ISP) to the internet. The internet is designed like a large road system—the routers keep deciding which road to take to get to grandma’s house. Eventually there are no turns left and the box is placed in her mailbox (i.e., her computer). The computer waits until all the packets have been received and then reassembles them back into the picture of a cat which is now on Granny’s screen.
TCP/IP is also fault-tolerant. During a session, it checks to make sure all packets are received and continually verifies the path. If, for some reason, a router along the path to the destination stops functioning, TCP/IP will automatically look for (and, most often, find) a different path to continue to deliver the data. This is why the internet never completely goes “down” as a whole; parts of it may stop working, but the rest of this vast network continues functioning without those parts.