Sharing Information Across the Office or Around the World

Ethernet is an inexpensive LAN (Local Area Network) technology used for transporting information from one location to another. In its largest context it is the backbone upon which the Internet and its World Wide Web (the quintessential WANs (Wide Area Networks)) are based, and in its smallest it can enable you to connect the computer in your child's bedroom to the one in your family room. Ethernet was developed in 1972 by two Engineers: Bob Metcalfe and D.R. Boggs. By 1980 it had caught on to the point that industry standards were established, known as IEEE 802.3. These standards define the protocols used to implement Ethernet communication, and they provide the guidelines used for manufacturing the computer hardware used to implement it--Ethernet cards and cables. Today, Ethernet technology is extremely common and $20 at a local computer store will get you an Ethernet card for your PC that will enable it to communicate with any other Ethernet enabled computer on the planet--(provided that computer is connected to the Internet or some other mutually accessible WAN).

Ethernet Technology

Like the traditional PC expansion choices such as PCI and PCMCIA, Ethernet uses a Bus topology. This means that all the devices on the network share a single communication line. However, when looking at a bus like PCI, that communication line is the system's CPU, for Ethernet it is a LAN made up of many Ethernet enabled machines. Addressing on an Ethernet LAN is very similar to addressing on a PC, but rather then COM ports, MAC (Media Access Control) Addresses are used to identify the proper message recipient. And while a PC with PCI-based serial ports may use the RS-232 serial protocol to process data, the Ethernet bus uses networking protocols to govern data processing, the most popular of which is IP (Internet Protocol. (See below for a discussion of networking protocols.))

Ethernet hardware is relatively simple and inexpensive. A NIC (Network Interface Card) is installed in a PC that directly interfaces with a computer's system bus. That card provides a connector for an Ethernet cable that plugged into the device itself and into one additional port on the LAN or WAN. The cables that make this connection are the major difference between the several types of Ethernet connections. The oldest Ethernet incarnation (10Base5) is often called Thicknet. It got this name because the cables required to make the connection were 10 millimeters thick. The wire used for ThickNet were coaxial cables (cables with an interior copper wire that carries the signal, then a layer of insulation, and another circle of copper mesh to act as a ground, and finally exterior insulation.). 10Base2, or Thinnet, was the next generation of coaxial cable. It provided much more cabling flexibility because it was only 5 millimeters thick.

A major advance in Ethernet cabling came with 10BaseT, still widely implemented today. The main advantage of 10BaseT is that it uses unshielded twisted pair (UTP) wiring rather than Coaxial cable. This not only enhances signal integrity, it also makes for much more economical implementation. Another cabling option is Fiber Optic, which can be implemented using 10Base-FP.

Another important hardware aspect to Ethernet is the maximum cable lengths permitted between devices, called a "segment." Each Ethernet segment can only be so long before the signals begin to fade or get corrupted due to line noise. The maximum distance for 10Base5 is 500m, 185m for 10Base2, and 100m for 10BaseT. Communication speed on these networks was also limited to 10 Mbits/sec (1.25 Mbytes/sec).

Today's Ethernet provides vastly improved data rates. 100BaseT uses UTP cable to provide data rates up to 100 Mbits/sec (12.5 Mbytes/sec). The similar 100Base-FX achieves the same data rates using Fiber Optic cable. And, both provide the added bonus of backward compatibility, meaning that both 10BaseT and 100BaseT cables can be used on the same networks. The newer Gigabit Ethernet can transmit data at speeds up to 1 Gigabit/sec (125 Mbytes/sec). Not widely used today, Gigabit Ethernet is catching on fast.

Ethernet Repeaters and Routers

Ethernet would be of little value if it was truly limited by the 100 meter segment requirement. To overcome this shortcoming, repeaters are used. The repeater functions exactly as its name suggests--it repeats and amplifies any signal that passes through it. It is essentially a dumb box, meaning that it does nothing but take data in at one end then repeat it louder out the other.

Routers, on the other hand, are actually computers whose only purpose is to connect networks together. The router functions much as does a Policeman directing traffic--based on the driver's destination, the router sends him on the most direct route to his desired location, and prevents other cars from hitting him or otherwise getting in his way. The router accomplishes this task by following predetermined network protocols for addressing computers--such as the IP (Internet Protocol) for directing data on the Internet (See below for more about networking protocols and IP addressing.). Often times Routers connected directly to a LAN will incorporate a Firewall, which ensures that only authorized data gets onto or out of the LAN.

Routers work together to keep traffic flowing over the network. Upon receiving a data packet, a router will determine the best route for the data to take in order to reach its destination. It will then address that packet to the next router in the optimal transmission line. This is called a "hop." Upon receiving the packet, that router will again look at the network to make sure that the packet goes to the next "hop" and address the packet to the best router, and so on, until the packet reaches its final destination.

Networking and Protocol Suites

In order to enable the worldwide network of LANs and routers to function properly, every computer on the network must adhere to the same set of rules. Networking itself follows the basics of data communication, Point A sends a message and Point B receives it. However, unlike the direct communication that takes place when a peripheral such as a printer is attached to a port on the PC itself, with networked devices there are a large number of middlemen required to get the message from Point A (the computer) to Point B (the printer). Not surprisingly, the more points there are between A and B, the more careful you need to be with the rules for passing the information, and the more fastidious you need to be with checking for errors.

There are two types of network that are used to transmit information--a LAN and a WAN. A LAN is a Local Area Network, which means a group of computers located in the same place--such as an office or a building-- that are connected via wired or wireless Ethernet. A WAN (Wide Area Network) is made up of many LANs that can be located any distance from each other. The Connections between the individual LANs that make up the WAN are made via a series of Routers that are all connected to the global network of communication lines. (This is made up of analog phone line, ISDN (Integrated Services Digital Network) or digital phone lines, the fiber optic cables run under the oceans, Cell towers, and even Satellite communications). Each individual computer on the network has its own address, and each Router is programmed to understand that address so that it can direct a message to the correct recipient. In the more familiar terms of the Internet, each ISP (Internet service provider) has at least one Router connected to the global network, and many large companies have their own routers that also connect directly to the Internet. Users not directly connected to a LAN with a router, can use a modem to connect with another modem that is part of a LAN which in turn translates the analog modem data into digital data and transmits it via its router onto the Internet WAN, which in turn gets it to the appropriate destination LAN, which many now translate it back to analog and transmit it via modem to another user not directly connected to a LAN.

With so many different kinds of communication going on at once over such great distances, it truly is a wonder that a PC in Antarctica can send a message to a computer in Alaska in a matter of minutes. However, to make it all work, there needs to be a strict set of rules that all components of the network follow, and there needs to be a way to quickly identify and correct corrupt transfers. The rules are called a Communication Protocol. There are many types of protocols designed to address the many different types of data transfer--some more familiar protocols are HTTP (used to transfer web pages) SMPT (used to send email), and FTP (used to send binary data files). To make things easier, multiple protocols used on the same type of network are grouped together as a "Protocol Suite." We will discuss the Internet Protocol Suite below, as it is the most common example. However, any group could create its own WAN that uses a completely different set of protocols or Protocol Suite, so long as they were agreed upon and adhered to by everyone on that WAN.

Protocol Suites are typically divided into layers that describe the different responsibilities of the hardware and software components that enable the network to function. The Internet Protocol Suite is based on the OSI seven layer model developed by the International Organization for Standardization. Each protocol in the Internet Protocol Suite corresponds to one or more OSI layers. The chart below defines the OSI layers and their corresponding Internet Protocol Suite components. The following table defines the different protocols that are part of the Suite.

Ethernet Frames and Collisions

Much like other protocols such as asynchronous and synchronous communication, Ethernet protocols are sent using a series of frames. An Ethernet frame is shown below:

An Ethernet frame is no more than 1518 bytes, and can be less depending on the size of the data packet. A header section comprised of a Preamble (indicating that a frame is about to be sent), a SOF (Start Of Frame) delimiter indicating that the preamble is over and that the transmission is about to begin, and a Destination Address (the destination IP) and Source Address (source IP). The following "Type" segment indicates the start of the data in the next frame, then comes the data itself, and finally the footer is a Frame Check Sequence which contains a CRC (cyclic redundancy check) value used to make sure that the data is not corrupted. The sending device creates the value based on the data, and the receiving end does the same calculation to ensure that it gets the same result as the sender. If not, the data must be retransmitted.

Data corruption is not the only peril facing ethernet transmissions, data collision is another. As might be expected, with so many devices attempting to talk on the network, two often try to do it at the same time. When this happens, a collision occurs. In order to avoid this, a protocol called CSMA/CD (Carrier Sense Multiple Access/Collision Detection) is used. It is a method by which any device wishing to transmit data first checks the lines to make sure it is clear, and then sends only if the lines are clear. This is exactly the same process used for other types of serial communication, such as RS-232 were the RTS (Request to Send) is transmitted and the CTS (clear to send) response must be received before the transmission can begin. However, with the large number of devices connected to an Ethernet Network, multiple devices may be listening at the same time, and though the line is clear when they listen, they all simultaneously transmit data thus causing a collision. When this happens, all of the involved transmissions fail, and a special algorithm is used by the network adapter to generate a random waiting period before trying the transmission again. Theoretically each device involved in a collision will generate a different delay, so that all of the re-transmissions will be successful.

Ethernet for Serial Communication

Ethernet uses different protocols than does serial communication, and unlike serial devices Ethernet devices are used to move data not acquire it. Thus, while you can ethernet enable your Printer (meaning you use an ethernet hub with a parallel port to which you connect the printer) you will never have and ethernet-based printer, one that plugs directly into the Ethernet port (unless the printer itself contains an Ethernet hub--and such devices are rare and expensive.) However, there are great advantages to being able to access devices that were traditionally accessible only from a single PC.

A typical such scenario uses a serial device such as a thermostat control in a remote warehouse. Traditionally that device would be connected directly to a PC running application software that uses COM 3 to transmit temperature adjustments. Unfortunately, anyone who wants to change the temperature in the warehouse must go to the local PC to do it. This can be very time consuming and expensive if that warehouse is located 100 miles from the nearest town. Using an Ethernet-Based Serial Device Server eliminates the need for a person to ever touch the local PC. Anyone can install the temperature monitoring application on a remote PC, and install drivers for the Serial Device Server that will enable them to connect to COM 3 on their local machine and actually transmit data to COM 3 on the serial device server to which the thermostat control is connected.

A serial device server, such as Quatech's ThinQ line, allows individual serial devices such as printers, simple terminals, medical monitoring equipment, CNC machines, etc., that were previously accessible only via a direct link to be accessible from any Internet enabled computer. According to Dataquest, a device server is a "specialized network-based hardware device designed to perform a single or specialized set of functions with client access independent of any operating system or proprietary protocol." In practical terms, this means that by using a serial device server you can connect any serial device to your network by connecting the device to a serial port on the SDS and connecting the Ethernet port on your SDS to your network. Once the connection is made, any PC or dumb terminal can transparently access the device as if it were accessing a native COM port.

While implementing serial devices via Ethernet does not provide any speed advantages over other busses (the standard data rate for serial communication is 115.2 kbps, and using special clocking techniques Quatech products for most bus options can go as fast as 921.6 kbps) it does provide far more flexibility. Devices no longer need to be located close to the PC itself--ideal for POS implementation, building access and security monitoring, and industrial CNC lines. In addition, commonly used devices such as printers can be easily shared among everyone on the network. And, using Ethernet to enable serial devices can actually result in more reliable communication because direct serial communication lacks the advanced error checking built into Ethernet frames.

Ethernet 10BaseT Specs
Bus Clock Signal		n/a
Bus Width		n/a
Theoretical Max. Transfer Rate		1.25 Mbyte/sec (10 Mbits/sec)
Advantages		Enables multiple PCs to remotely share information and peripheral devices, provides error checking lacking in standard serial communication
Disadvantages		Most peripheral devices cannot be connected directly to ethernet, adapter is required. Slow data communication by current standards Possible security issues.

Click here to see how Ethernet 10BaseT compares with other busses.

Ethernet 100BaseT (Fast Ethernet) Specs
Bus Clock Signal		n/a
Bus Width		n/a
Theoretical Max. Transfer Rate		12.5 MBytes/sec (100 Mbits/sec)
Advantages		All advantages of 10BaseT, with significant speed improvement. Backward compatible with 10Base T installations.
Disadvantages		Most peripheral devices cannot be connected directly to ethernet, adapter is required. Possible security issues.

Click here to see how Ethernet 100BaseT (Fast Ethernet) compares with other busses.