Fundamentals of Communications

This segment is going to cover the communications basics necessary for the rest of this series.

We've divided this segment into three parts. The first is the Public Switched Telephone Network (PSTN.) The second is the Data Switching Fundamentals, and the third is the Public Data Network, or PDN.

The first of these, the Public Switched Telephone Network, or POTS (which stands for Plain Old Telephone System) is a very important component of developing a strategy for future digital networks because this is the structure that is in place and does exist and we're going to be relying upon for many years. This part of the presentation is divided into four parts. The first is divestiture; the second is some terminology; the third is deregulation; and the fourth will cover standards.

In 1982, Judge Greene issued the Modified Final Judgment, which broke the long monopoly that AT&T had had for providing telephone service in the United States. AT&T retained the ability to provide long-distance service, the right to manufacture telecommunications systems, and was allowed for the first time to enter the data processing arena. The regional Bell Holding Companies, or RBHC's, such as US West or Bell Atlantic, continued to provide local telephone service as a regulated monopoly. They also were able to provide new, unregulated directory services.

It's important to understand the distinction between local and long distance after divestiture occurred in 1982. The country was divided into LATA's, which stands for Local Access and Transport Areas. These LATA's roughly correspond to the area code in a telephone number. Within an area code, the local exchange carrier, or LEC, has, until very recently, a monopoly of providing the telephone on your desktop. However, the local exchange carrier cannot connect telephones in two different LATA's. The LEC must contract with an IXC, or Inter Exchange Carrier, to connect two different LATA's. Two other definitions will be useful in this series. One is for Centrex. Centrex is the technology that we're familiar with on our desktop. This is the telephone service provided by the telephone company. This is what we have in our homes. In addition, many institutions build their own telephone system. This is called a PBX, or Private Branch Exchange. Many of our campuses operate their own telephone systems on campus. This is the PBX.

There are a number of standards that we need to be familiar with as we proceed through this series of presentations. The first is for Twisted Pair. Twisted pair is a very simple concept. We're going to take two wires and we're going to twist them around each other. Why do we twist the wires around each other? The reason is that the wire itself acts as a radio antenna. It both broadcasts and receives. By twisting the wires together, we reduce the interference from other signals.

There are a number of categories of twisted pair that will be important. The first is Category 1. This is essentially a set of standards that defines the minimum needed to transmit voice. Category 3 is the one that is most common in our older installations, and that is capable of supporting voice and medium speed data. The final category is Category 5. Category 5 is a recent standard and it is capable of supporting voice and what, in the telecommunications industry, is regarded as high speed data. In our environment, that might be considered to be medium speed data, but Category 5 wire will support data transmission speeds of 100 to 150 megabit per second.

As we go to a larger scale, there are some transmission standards that you need to understand. When you pick up a telephone, you have an amount of band width that roughly corresponds to four kHz (kilohertz). That signal is an analog signal. As it goes upstream in the telephone system, that analog signal is digitized. It is digitized into seven bits of data and an additional bit is included for a control signal. To adequately digitize a four kHz analog signal, you need to sample that signal 8,000 times per second. So, if we take eight bits per sample times 8,000 samples per second, we have a 64 kilobit or 64k data stream. That data stream now consists of 56k of data and 8k of control signal. So your voice signal started out as a 4kHz analog signal and it is now residing in a 6kHz digital signal. Twenty-four of those 64kHz digital signals are now combined into a 1.54 mbps trunk that is usually referred to as a DS1 or a T1. There is a subtle distinction between DS1 and T1, but for the purposes of this presentation, we are going to ignore them. Once again, we can combine twenty-four of these DS1 streams into a single stream of 45 mb, which is referred to as a DS3 or T3.

In this segment, I'd like to review some data networking fundamentals. We're going to cover four basic areas. The first is Local Area Networks, or LAN's; the second one is Metropolitan Area Networks, or MAN's; the third is Wide-Area Networks, or WAN's. And finally, I'd like to review some basic concepts that will be useful for the rest of the series.

Local Area Networks, or LAN's. These are networks that are typically rather small. They may be a few meters to a kilometer in size. This might represent a building or perhaps a small campus. Within this small area, it's relatively easy to achieve transmission speeds of megabits per second, so within this area, typical speeds would be 5 Mbps, 10 Mbps, 16 Mbps, or even 100 Mbps. Understanding Local Area Networks is actually much easier than it appears to be on the surface.

There are four basic criteria, four basic parameters that will allow us to categorize almost any Local Area Network. The first of these parameters is the media; the second is the topology; the third is the signaling method, and the fourth is the access mechanism.

The first of these is the media. What is the physical thing that we send the signal on? The media might be a radio wave, for example; a microwave. It might be a lower-frequency radio wave, AM or FM. It could be anything that transmits the signal. For the purpose of this series, we're really going to look at three transmission media in detail. The first is twisted pair. Twisted pair, again, is just two pieces of wire that have been twisted together. Typically, the maximum data rate for twisted pair is measured in a few tens of megabits per second. 10 Mbps, 16 Mbps, or perhaps even as much as 100 Mbps. Typically, there's a maximum range on this twisted pair of one kilometer. The advantage of twisted pair is that it is very inexpensive. A lot of wire already exists. Sometimes we can make use of it; sometimes we can't. But even if we install new wire, it's a technology that we're very familiar with, our technicians know how to install it, and the cable itself is relatively inexpensive. The drawback to twisted pair, as I indicated earlier, is the fact that it is susceptible to electronic noise. The wire acts as an antenna, both to transmit a radio signal and to receive radio signals, and these signals can interfere with out digital transmissions.

A second media that is frequently found on our campuses is coaxial cable. Now, you're familiar with coaxial cable because that's the stuff that goes into your cable television set. It's a center- conductor, surrounded by an insulator and then surrounded by a circular jacket through the entire length of the cable. This cable typically has a maximum data rate somewhat higher than twisted pair. It can easily range into several hundred megabits per second. Typically, the maximum range is about the same, about one kilometer. It is slightly more expensive to install than twisted pair, but it is relatively immune to electronic noise.

Optical fiber is becoming an increasingly popular transmission media. How does fiber optics work? As you recall from your high school science courses, as light moves between a more-dense and less- dense media, it is bent. And if the angle from a more-dense media moving into a less-dense media is sufficiently shallow, all of the light is reflected internally, and none of it is transmitted into the less-dense media. So, if a light signal is sent down a very, very thin piece of glass -- a piece of glass that resembles a very thin string -- that light will be reflected internally, and none of it will be transmitted to the outside. And so the light signal simply bounces down the walls of this light pipe. Now, a fiber optic cable provides a very high band width. It's very immune to electronic noise because the electromagnetic signal isn't in the radio frequency ranges at all; it's simply a light signal. It is a point-to-point mechanism; it's extremely reliable. If it works, it works. And if it doesn't work, it doesn't work completely; it's a piece of glass. It's broken. And it's straight forward to fix the one single break. It is very difficult to tap, so it provides a very high degree of security.

There are two kinds of fiber commonly used on our campuses. One is multi-mode, and one is single mode. Multi-mode fiber typically has a maximum range of a few kilometers. Now, multi-mode fiber, if you were to look at it, would seem quite thin. It's about the thickness of a piece of thread. However, even as thin as a piece of thread, it's possible for light rays as they reflect down this pipe to take different paths. For example, one light ray might come in at very shallow angles and go down the pipe, ping! ping! ping! Another light ray might come into the pipe at a somewhat steeper angle, and it would reflect down the pipe, pingpingpingpingping. The result is that those two light rays will get to the end in slightly different times. So, a digital signal will gradually be dispersed and spread out, and so, after some distance, there has to be a receiver which looks at that light signal (which now becomes kind smeared out), digitize it once again and re-transmit it as a clean signal. Single-mode fiber is simply a thinner piece of fiber. Now we're talking about fiber that is, perhaps, the thickness of a human hair. Now that increases the range of the fiber, because there is less dispersion. The packets that go down the fiber all reflect about the same way, so the transmission distance is longer before it has to be regenerated. On the other hand, the fiber itself is more expensive to manufacture, and it's much more difficult to work with.

There are four commonly-used media on our campuses. One is 10Base5. We're familiar with this as the old thick coax that we used for Ethernet. It has a typical maximum length of 500 meters and can support a number of users. There is also what is called thin coax, 10Base2. Its range is somewhat shorter, but it's much easier to work with. You should have noticed that the nomenclature is that the first number represents the speed of transmission, then base for "baseband", and then a number that represents how many hundred meters the signal can be transferred. So 10Base5 refers to a media that can operate at 10 Mbps for 500 meters. As we look at the third of these media, we see 10BaseT. This is twisted pair designed to support a 10 Mb data stream with a distance limitation of 100 meters. That 100-meter limitation will become very important later in the presentation. And finally, I have identified 10BaseF, which is a fiber optic cable capable of supporting a 10 Mb data stream for 2,000 meters.

The second parameter that we use in characterizing Local Area Networks, or LAN's, is topology. The star topology has a central hub with direct connections to distributed nodes. A ring topology, on the other hand, connects the nodes in a sequential fashion so that a message would pass from one node to the next to the next, and back to the original node. In a bus topology, a message is broadcast and all of the nodes see the same set of messages at roughly the same time. There are variations on these topologies. For example, one variation on the star is a tree topology. There, a central hub sends messages to distributed nodes by individual media. Each of those nodes, in turn, may distribute to additional nodes through, again, a star configuration. This tree is the kind of topology that is used, for example, in a cable television signal.

Signaling method is another way that we can differentiate between various kinds of Local Area Networks. We'll be concerned primarily with two signaling methods; baseband and broadband. I should add that the term broadband will be used in a different context as we talk about high performance Wide Area Networks. A baseband signaling technology makes use of a single data channel. In other words, an electrical voltage or a light signal is placed on the media. A good example would be a telegraph, in which an electrical voltage is placed on a wire. Thus, you would have a series of zeroes and ones --da di da dit -- da da di da. Those signals would either be zero or one, and there is only one data channel. In a broadband technology, on the other hand, we have multiple data channels, so we might have, for example, multiple communications being transmitted over the same piece of media. This can be done by stacking these various data channels in terms of their frequency. For example, a television signal requires about two megahertz of band width. We might, then, take that two MHz signal and add 100 MHz to it. Then we would have a signal ranging between 100 and 102 MHz. We might take another channel and add 102 MHz to it, so that television channel would range between 102 and 104 MHz. We might take a third television channel, which again requires two MHz of band width, and add 104 MHz to it. Thus, all three signals would reside in the 104 to 106 MHz range. The job of the television tuner is then to select one of those ranges, subtract from that frequency the base amount that was added, to extract the original signal. What we have done is to take multiple data channels and send it simultaneously over the same media. The disadvantage of this technology is that it requires active RF devices at each end of the signal.

The final criteria that we use for differentiating between various kinds of LAN's is the access mechanism. Essentially, we want to know who determines who is talking and who is listening. There are two commonly used access mechanisms. One is CSMA/CD, which is used in Ethernet, and the other is Token Passing. CSMA/CD is a complicated term. It stands for Carrier-Sense Multiple Access/Collision Detection. Essentially, it is a polite, cocktail-party mode of conversation. In polite cocktail conversation, we're in a group, we have multiple access (because anyone can talk) and we assume that everyone is listening, if we have something to say, we wait. We listen. Is anyone else talking? That's the Carrier Sense. If no one else is talking, then we say what we have to say. However, we all know that what frequently happens is that someone else will decide to say something at the same time. What happens? Well, two people start talking at the same time. They stop. They back off some random length of time, and they try again. So there was Collision Detection -- we backed off a random length of time -- and we re-transmit. Now, sometimes you get a second collision. The two people start talking at the same time again. This time, they blush slightly and they back off a little bit longer, and, again, a random length of time determined by the person's personality. And perhaps there's a third collision. At this point, everyone collapses into giggles and the random back-off is even more random and even longer. This mode of access is very statistical; it's also very simple. It does not require any centralized management schemes. Token Passing, on the other hand, is much more deterministic. We might take our cocktail party model and have everyone arranged in a circle and take a card. And the card is passed around the circle, and the only time you can talk is when you have the card in your hand. So you wait patiently. When you have the card, you say what you have to say, and then you give the card to your neighbor. This deterministic mechanism requires much more management. It's much more complex. However, it works better under high traffic-load situations.

Another major area that we're going to talk about briefly are Metropolitan Area Networks, or MAN's. Here, we're talking about inter-processor distances from ten kilometers, 15 kilometers, 20 kilometers, five kilometers -- something that might represent a big campus, a small city, or metropolitan area. One of the most popular technologies makes use of one or two cables and has no active switching elements. One of the terms that you may run into is DQDB, which stands for Distributed Queue Dual Bus. We're not going to talk about this protocol in any kind of detail, but you should simply recognize the term.

Now we come to Wide-Area Networks. Here, we're talking about networks that span hundreds to thousands of kilometers. How do we connect our campuses to other campuses? How do we connect one city to another city? How do we connect one continent to another continent? Here, we're forced to rely upon either the plain old telephone system or the public data network. As an individual campus, I can't run cable between my campus and a campus on the other coast. I can't afford to trench for thousands of miles. I don't have the right-of-way, nor could I afford it. So we have to make use of the plain old telephone system, or the public data network, which is the infrastructure that the telecommunications companies have placed in the ground to support digital communications.

One set of definitions that will be useful as we talk about networks is a layered architecture. This can become quite complex, but let me try to explain it in very simple terms. In early days of computer programming, someone might write a computer program -- let's say, a spreadsheet application. Well, they wrote the spreadsheet application, but then they realized that they had to write a program to light up the CRT tube, so they wrote that in their computer code. They had to write another program to be able to read the signal from the keyboard. And they wrote other programs that would run a printer. When they wrote that code, they wrote this as one, big enchilada -- everything was mixed together. And it worked. But then somebody changed the printer. That meant that the author had to go back to thousands of lines of code and change a line here, and line there, to deal with the change in printers. And someone changes a keyboard, and again, the author had to go back through thousands of lines of code and change a line here, and a line there. What evolved was a set of architectures in which very clear distinctions are made between what each portion of the code does. In the case of a network, we have an architecture that says there are applications at the highest level. In the ISO model, that's referred to as Level 7. I don't want to go into a lot of details as to what these levels do, but you will hear them referred to frequently.

At the top levels, we see the applications. We see the instructions that drive the computer screen. We see the instructions that are able to access data from other programs and move them between the programs. As we go down the stack, to the very bottom end, we see the physical layer. And this is a set of standards, that define a physical connection. We may arbitrarily say that the connector will have this shape, it will have so many pins, that the signal will be electrical, and a logical one may range between 3.3 volts and 5.6 volts. So it's very specific and very detailed. But it doesn't say anything about our application. Slightly above that, there's a layer that tries to look at how this information might be organized into a packet to be sent across a network. And still higher, we see the network layer, which in TCP/IP would correspond to IP. How do we organize this information in packets that are useful? How do we put the headers on it? The next layer up, the transport layer -- which in TCP/IP is TCP -- this tries to determine how we get the packet of information from one point to another point. What route do we take, how does it arrive, and so on. That segment then communicates with the layer above it, which moves on to the applications layer. The communication between two different computing devices always occurs at the bottom layer. It's a physical signal. It's either a light pulse in a piece of fiber optics, or it's a voltage on a piece of wire. So transmission between the applications does not go from application to application. The application on one computer starts talking to layers below it until it finally reaches the physical layer. Then a signal is sent through the network over to another physical layer, and then back up the stack to the other application.

There are some other distinctions that will be useful in the course of this series. The first is the distinction between packet switching and circuit switching. Let's take the second of these first. Circuit switching establishes a dedicated transmission channel between the two users. For example, I might choose to communicate with Joe using a tin can telephone. And to do that, we each have a tin can and we have a piece of string between our two cans. In the case of a telephone system, we might pull a piece of wire between myself and Joe. Now we can add variations to this circuit switching environment, and we might say it's too expensive for me to have a dedicated piece of wire to everyone I might talk to, so I'll put in some switches in this process. That's what we do in the case of the telephone system. I have a line to my local telephone office, which is connected to a switch, and I can establish through that series of switches a link to whomever I wish to communicate. It is a dial system, because I can set up the circuit to anyone in the world by dialing a number. Now, if I'm frequently connecting to the same people, I may want to set up a cross-connect. That's sort of a default dialed line, which I can have changed, but generally stays with the same circuit. And finally, I may tell the telephone company, "I would like to lease a private line. I want to have a physical connection between my site and another site, and I want it to be available 24 hours a day, and operational 24 hours a day. And I don't want to share it with anybody!" Packet switching, on the other hand, is a store-and-forward technology where a signal is sent from one node to another. Think of this as taking a message, "To be or not to be," and I might break that message into three pieces; "to be," a second, "or," and a third, "to be." Take each of those messages and put them in an envelope, and address each of those envelopes and put them in the mail, Now, the recipient will receive those messages some days later. Will he receive all of them at the same time? Probably not. Will they all have gone by the same route? Not necessarily. We put them in the mail; one might have been transmitted via St. Louis, another via Minneapolis. One might arrive in two days, one might arrive in three days, one might never arrive. So we've broken the message into packets and we've sent these packets out into the communication system. And each packet has an address. In this kind of environment, we can actually make an additional distinction; and that is between connectionless and connection-oriented circuits. In a connectionless oriented service, sometimes called a datagram, each packet is independently routed. In other words, we took our message, "to be or not to be." We broke it into three packets. We put them in three separate envelopes. Each envelope had an independent address, and, as these went through the mail system, they might or might not follow the same path. It would be determined by how many letters were on the truck or the train or the airplane. In a connection-oriented service, sometimes called a virtual circuit, the first packet defines the path for all of the following packets. So the first packet through creates a route that all following packets in that message will follow.

Quality of service is a technical term that is becoming increasingly important as we migrate to advanced networking technologies. Quality of service refers to delay and jitter. Delay is the time it takes a packet or a cell to go from its source to its destination, and jitter is the variation in that delay. Quality of service, delay, and jitter is essential for the real-time transmission of voice and video. As I previously indicated, voice does not require a great deal of band width. However, the human ear is extremely sensitive to the timing of the packets as they arrive. If a sig naliskindof brok enupand th epac kage arriveit'sa lit tleout oftime sequ ence, it's very hard to understand the signal. A video signal imposes the same kind of time constraints as well as requiring more band width. So quality of service is the essential ingredient for transmitting real-time voice and video. This is commonly done today with two technologies. One is ATM, which will be discussed later, and the other is the next generation of TCP/IP, called IP Version6 (IPv6), which makes use of RSVP. This will also be discussed later in the series.

I'd like to focus our discussion of public data networks on five main points; the role of Wide Area Networks, the types of service that are available, two older technologies (X.25 and ISDN), and finally, Fast Packet Switching. What is the role of the public data network in our campus communications? As a campus, we can establish a digital network that will provide the services that we desire. However, to connect to other campuses and other institutions, we have to rely upon the public data network provided by the telecommunication carriers.

As we consider types of service, I would like to make two sets of distinctions. The first is between dedicated and switched. For example, in a dedicated service, we might choose to lease a voice-grade telephone line. We might also lease a higher band width line such as DS1 or DS3, or perhaps even SONET, which we'll discuss later. Finally, we might choose to lease a piece of dark fiber. Dark fiber is simply a piece of fiber optics without the electronics at either end that would send a light signal down the fiber. Or I might elect to make use of a switched service. Telephone circuits, for example, are switched service. This would include our analog and digital telephone systems. We pick up the telephone headset, we dial a number. That signal is then switched to wherever it is we have selected. We could also obtain switched 56 kb or fractional T1 service, so we could elect a switched 56k service or some sub-multiple of a T1. Finally, ISDN and SMDS Frame Relay and Cell Relay are all switched services that we'll be considering shortly.

The second set of distinctions that I'd like to make is between circuit switched and packet switched. We've alluded to this distinction already. A telephone circuit is circuit switched. We actually establish a dedicated link between the source and the destination, and we have a series of switches that establish this link. Typically, the kind of switched services that are available are 56 kb and fractional DS1. This would be multiples of 56 kb up to a 1.5 mb speed. In a packet-switched environment, we can make two further distinctions. We can distinguish between connection-oriented services, such as Frame Relay and Cell Relay, and connectionless oriented services such as ISDN and SMDS. In a packet switched service, remember, we are taking the information and breaking it into discrete packets. In the connection-oriented service, we're letting the first packet through the system define the route. In a connectionless oriented service, each packet is independently addressed.

I would like to briefly comment on X.25, an older digital service offered by the telecommunications companies. It is a connection oriented virtual circuit. There are two distinctions that are important. One is between a switchable virtual circuit, referred to as an SVC, and a permanent virtual circuit, referred to as a PVC. Those are two terms that you will run into again. An X.25 connection assumes that there is a very high data rate in the transmission media, and that there are dumb devices at each end. In other words, it was based on an older generation of copper technology in which there was a high error rate, and it was developed at a time when we didn't have microcomputers on everyone's desk. At best, we might have a dumb terminal. As a result of this high error rate and having dumb devices at each end, the network had to assume the responsibility of correcting for errors between each link in the network. So the network itself had to be much smarter. Typically, the data speeds available were 64 kbps. The fundamental assumptions underlying an X.25 network, however, are no longer true. Error rates on our optical networks have fallen dramatically, and we can now assume that users have intelligent devices on their desks. So X.25 was a transition technology between our older, analog telephone environment and our newer, all digital environments based on fiber optic cross country lines.

ISDN, or Integrated Services Digital Network, is a service that is now becoming widely available throughout this country. It is a connectionless, or datagram service. In other words, each packet carries an independent address. The basic interface standard is two B plus one D channels. Each B channel is 64 kb and can carry either voice or data. There is also a 16 kb signaling channel. So an ISDN connection into my home could be configured to support one voice channel of 64 kb and one data channel of 64 kb, or I could choose to configure that system so that I have one 128 data channel and I won't use the telephone. Because of the connectionless oriented nature of ISDN, it is most suitable for voice and data, but it doesn't work very well as we try to deal with higher speed transmissions or real-time video.

What is the role and future of ISDN? ISDN is probably a day late and a dollar short. It does not meet our future requirements. It does, however, probably have a significant role as we try to extend connectivity into the home. So, for a number of years, ISDN may prove the best way to provide connectivity to our faculty, our students and our researchers in their home.

The final area of the Public Data Network that I'd like to consider is Fast Packet Switching. Fast packet switching is a concept, not a technology. It makes a number of key assumptions. First, it assumes that there are high speed transmission links. These would be the fiber optic SONET connections. It assumes that there is a very low error rate on these links. This is one characteristic of the fiber optic transmission network that has been installed throughout the country. Third, it assumes that there are intelligent devices at each end of the connection. These intelligent devices can then do much of the error control and flow control that used to be done by the network itself. And fourth, fast packet switching really represents a streamlined version of our conventional packet switching (such as X.25) that minimizes error checking and flow control.

Fast packet switching can conceptually be broken into two parts. One is Frame Relay and one is Cell Relay. I'll talk about both of those in just a moment. Frame Relay has a series of standards, and in the far right hand side we see that it offers a service. One of those services is conventional ISDN, and another service is Frame Relay. As we go to the lower branch of the chart, we see Cell Relay. In the case of Cell Relay, the standard that is used is ATM and the service provided is SMDS. This is going to create a little bit of confusion later on.

Let's take a look at Frame Relay in a little more detail. Frame Relay is a connection oriented service; that is, the first packet establishes the routing for subsequent packets in the message. It establishes either a switched virtual circuit or a permanent virtual circuit. Frame Relay is good for medium speed data service. The frames within the packets have a variable length. This allows Frame Relay to make very efficient use of network band width. However, it also introduces some variable delays, which can lead to problems with the transmission of voice and precludes the use of Frame Relay for real-time video. Frame Relay is a technology that operates up through level 2 of the OSI protocol stack that I referred to earlier. That means that Frame Relay does have some knowledge of the protocols and the languages that are being transmitted over the network, and that requires some technical adjustment.

Cell Relay is also a connection oriented service. It establishes a switched virtual circuit or a permanent virtual circuit. However, the length of each cell, or packet, is now fixed. And that means that the switch that switches this packet, or in this case, cell can be made to operate at extremely high speeds. Since each cell is a fixed length, the switch can design in silicon things that, for frame relay, had to be done in software. A cell relay system works through OSI level 1. Basically, that means a cell relay switch is very fast and very stupid. It's totally protocol-independent; it doesn't care what language is being transmitted. All it knows is it's going to transmit it very, very quickly. The delays are small enough in a Cell Relay system to support the real-time exchange of voice as well as video information. This, of course, makes Cell Relay very attractive to telecommunications carriers because using this technology, they would be able to support not only data transmissions, but voice transmissions and video transmissions as well. Within Cell Relay, there are two primary sets of standards and services.

The first is ATM. ATM is the underlying technology and broadband ISDN is the service that would be offered to the end consumer. Broadband ISDN is, again, based on ATM technology as well as a technology called SONET, which we'll talk about shortly. SMDS is a service that is based on the DQDB, the Distributed Queue Dual Bus technology that I briefly mentioned earlier, and finds its primary role in Metropolitan Area Networks.

The Switched Multi-Megabit Data Service, SMDS, it is connectionless. In other words, each packet has its own independent address. It was designed primary for Metropolitan Area Networks and establishing a logical private data network for a company that might have distributed sites. The standards that SMDS is based upon is the DQDB standard, and it makes use of DS1 or DS3 access transmission. Typically, an SMDS signal has a ten-millisecond packet delay, with 95% of the packets arriving within 20 milliseconds. This makes SMDS very well-suited for data transmission, but less well-suited for real-time voice or video.

Asynchronous Transfer Mode, or ATM is a connection-oriented circuit, and it forms the technical base, along with SONET, for a service called Broadband ISDN. And here, the word broadband does not refer to the switching mechanism we talked about earlier, but refers instead to this set of technologies. ATM is intended for high-speed Wide Area Networks. Typically, we see speeds of DS3 and OC3 and above. ATM is capable of supporting integrated voice, video, and data over a single media.

SONET stands for Synchronous Optical Network, and this is the transport service that is used in optical networks. In a SONET system, we multiplex and de-multiplex both synchronous and asynchronous cells and transmit them through the system. One of the advantages of SONET is that it can de- multiplex a portion of the overall transmission stream without de-multiplexing the entire stream. The basic building block of a SONET network is 51 Mbps. SONET has payload transparency; it is simply transmitting flashes of light down a piece of fiber optic.

The SONET optical line rates, as I said, were multiples of 51Mbps, so OC-1 would be one times 51 mbps; OC-3 is three times that, or roughly 155 Mbps; OC-12 a little over 660 Mbps, and OC-48 something like 2.5 Gbps.

The references for this series, if you would like to get more information about data networks, are Computer Networks by Tanenbaum and a publication by CAUSE on campus telephone systems titled "The Information Technology Manager's Guide to Campus Phone Operations."

We hope that you will find this technology primer useful in understanding the rest of this series.

About CREN

© CREN, 1999

Contact Us