The Internet as we know it today has grown in leaps and bounds from when it first really became popular back in 1994. Back then a 14.4K modem was pretty happening. Well times have changed and technology has certainly evolved since then. Presently most of us have high speed modems whether they be DSL or broadband cable modems. These modems transmit and receive data that your computer generates on your behalf. This is done in a very orderly fashion by your operating system. Seen as most of us use one version or another of Microsoft Windows we can map how our computer generates this information by mapping it to the OSI Reference Model. Almost each layer of this communications model has some information added to it by your computer’s operating system.

Let's think of this from the top on down. At the application layer Internet Explorer would make a HTTP GET request for your home page. There would be nothing for the presentation layer or session layer to add. Next your computer would generate a TCP header and the network layer an IP header. After that your Data Link Layer would put this information in a logical format and put the MAC information at the front of this frame. Lastly, it would be sent out via the NIC card at the physical layer. Bearing this trickling down of bytes from the application layer on down, we realize that each packet will likely not always be the same size. That is no problem for the computer, but it does present a problem if you want to switch this information around faster. The various shapes and sizes of Ethernet packets presents a challenge if we want to speed things up. Being predictable is not always a bad thing. Knowing that a computer will always generate a certain frame size will certainly help in efforts to design a faster way of switching these now uniform size packets.

Just like a pinball game

Clever folks realized that having packets generated by the operating system’s TCP/IP stack involved continually changing packets sizes. These engineers decided that it was time to design some new technologies to take advantage of faster networks. While this transmission of data may happen in a blink of an eye from our perspective it can certainly still be speeded up. Well, borne from this realization Frame Relay was designed and began to take off in terms of popularity and general acceptance. In reality Frame Relay has two components to it. That of the physical “layer one”, which is the physical interface of Frame Relay ie: RS-232. Also seen as Frame Relay is a switching technology it resides at the Data Link Layer which is “layer two”.

Now you could have a router configured to act as a Frame Relay switch and that allows for quicker communications as all that will change in it is the data segment, which can be of variable length. The remainder of it is the same.

8 bytes             16 bytes                       Variable                       16 bytes           8 bytes
Flags                 Address                         Data                              FCS               Flags

The above noted is what a Frame Relay “frame” looks like. We can see that the only part that changes in size is the data portion. In the first part is the flags field, which marks both the beginning and the end of the frame itself. After the flags field we have the address and it contains various pieces of data. A key piece of data contained in here is the DLCI. This value is used for identification purposes and in reality is how the network routes the data itself. It is also within the Address field that congestion values can be noted if experienced during transmission.

Next in the caboose is the data section and is then followed by the FCS (frame check sequence) field. This field is also rather important as it is used for data integrity purposes. Much like the checksum field in the TCP/IP core protocols this value is computed by the source device and is then recomputed by the destination device to confirm integrity. It stands to reason that you would want to know if your data has been corrupted somehow during transmission. Lastly, we have the flags section once again to signal the end of the frame itself. It should be noted that this is a standard frame relay frame and there exists another frame type. This would be the LMI frame format. It is significantly different and I would encourage you to give it a quick Google.

ATM isn’t only for banks!

A while ago a colleague and I were chatting about ATM (asynchronous transfer mode) when another colleague walked in. He started complaining about the lineup at his banks ATM. This gave us both a good laugh as we were talking about the switching technology vice the actual banking services offered via an ATM terminal. That being said a lot of people are still confused by this and swear that their ATM uses ATM! Needless to say not a lot of people outside of the networking world are aware of this switching technology. Over the years, as our connections to the Internet have grown quicker, few have actually given any thought to just how these connections have gotten quicker. Or more to the point, just how that internal work connection is so darn fast.

Much like the above discussed Frame Relay is the fact that ATM operates at the Data Link Layer in the OSI Reference model. ATM was borne out of the fact that high volume traffic needed to be sent and a better way was needed to do it. It will likely be a reality in our lifetimes that we will see the foundation for a true fiber optic network laid down by the major Internet service providers that will impact us as home users. With ATM to do a lot of the grunt work this will help this become a reality. Much like Frame Relay, ATM has hardware requirements that come into the picture. Hardware appliances such as multiplexers, switches, and routers are used in conjunction with ATM. Hardware devices are great, but as usual you still need a software front end to control them normally.

With the very impressive speed of ATM one has to wonder just how it is able to switch traffic about so quickly. Well unlike the random packet sizes of TCP/IP, ATM has a fixed cell size. This way the incoming traffic is always the same size and that makes things simpler for lack of a better term. That simplicity allows for the vastly increased data throughput. What does an ATM transmission look like though? Let’s take a look!

       <-----------5 Bytes---------->                    48 bytes
VCI Label     control     Header checksum          payload

The above noted 53 cell formation is the way it will look every time. Due to this predictability comes the ability to switch these cells very quickly. Much as we saw above with Frame Relay is the key word “predictability”. Both of these switching technologies allow for faster switching of data due to their predictable nature. This is done via their encapsulation of upper layer data in their respective formats. This was very much a high level overview of both Frame Relay and ATM. Hopefully it will be enough to whet your appetite to research them both more. As we are beginning to see all of the world is not composed of only TCP/IP.