Previous | Table of Contents | Next |
Connectionless switching is more suited to data services that require access of a bursty nature. Rather than tie up a physical channel that may remain idle for long periods of time, data services that transmit data in a pseudo random or bursty manner use connectionless switching. Their data is multiplexed with other bursty services. Network engineers, using connection probability theory, calculate the optimum number of connectionless switches/multiplexers required to support a network within the limits of some defined Quality of Service (QoS) parameters. To manage a bursty data stream through the network, various methods of grouping the data bits of each channel have been devised. Data stream management is necessary to ensure that the specified QoS is maintained. That is, all the bits, within acceptable limits, must get successfully from the source to the destination.
Packet switching hacks the data into segments. The data to be transmitted is divided into chunks or packets. Each packet contains two groups of bits. One group is the data and the other group of bits is the header information used for routing and quality control. At each node of the network the packet is received, stored briefly, then passed on to the next node. At each node the packets may be put on a queue to wait for an open channel before moving further into the network. There can be several nodes in the transmission path. There are two approaches used to transport the packets. A datagram is where each packet can take any path through the network as long as all the packets reach the destination, eventually and not necessarily in the order transmitted. Virtual circuit (VC) is where all the packets are routed through the same path without having the path dedicated. Datagram allows for dynamic handling of congestion (sends packets around the bottleneckyou take the high road, Ill take the low road kind of thing and we will meet in the morn) and no call setup is necessary. VC channels allow for sequencing (you get each packet in the same order I sent it), and error (bad puppy...uh packet...send it to me again) and flow control (slow down, I cannot read as fast as you can e-mail me).
Multirate circuit switching (MRCS) is an enhancement of the synchronous TDM approach used initially in circuit switching. In circuit switching, a node must operate at a fixed data rate which must be used regardless of application. In multirate switching, multiplexing is introduced. A node attaches to the network by means of a single physical link which carries multiple, fixed data rate channels called B-channels at 64 Kbps. Traffic on each channel can be switched independently through the network to various destinations. This is used for simple ISDN. The user has a limited number of data rate choices but they are fixed so variable bit rate (VBR) is difficult to accommodate efficiently. A user that desires to transmit MCS data at rates other than the CBRs must use a higher rate than the maximum predicted.
Since VBR has been introduced, a few comments about it is in order. VBR is a useful tool for controlling the transmission system loading, ensuring that during peak traffic periods the system resources, such as transmission facilities, do not get overloaded (you can only shove so many bytes down the throat at any one moment regardless of the ability to swallow), resulting in lost data.
Frame relay is essentially the same as packet switching. Frame relay was developed as a result of the high data rates desired combined with the need for low transmission error rates in modern high-speed data communications systems. In packet switching, there was considerable overhead involved in error recovery (leave me alone, I can fix it myself), redundancy enhancement (play it again, Sam, I didnt get it right the first time), and routing information (where, oh where, has my little dog gone? Where, oh where, can he be?). Overhead is synonymous with inefficiency. With frame relay, the packets are variable length, not fixed length, meaning that they are designed to operate at up to 2 Mbps. The variable length frame relay packets provide a means for implementing VBR.
Cell relay is an evolutionary descendant of frame relay and multirate circuit switching. Cell relay uses small, fixed-sized data packets called cells. The cell relay switching approach allows for the definition of virtual channels (multirate circuit switching has fixed channels) with data rates dynamically defined. A small cell size allows an almost constant data rate even though a packet is the transport structure. Cell relay error control is much improved over frame relay, and allows more errors to be handled at a higher logical level (more play it again, Sam ...but now the request comes from Louie). The small, fixed-size cells reduce overhead even more and thus allow rates of tens to hundreds of Mbps. Cell relay is equally adapted for use with variable bit rate (VBR) and constant bit rate (CBR) switching technologies.
We are now in possession of a fast switching technology, cell relay, and a fast transmission technology, SONET/SDH, over optical fiber. Cell relay supports VBR, CBR, and varied applications. SONET/SDH allows large amounts of data to be transmitted efficiently over a network. The combination of these two technologies, cell relay and SONET/SDH, provides the foundation for ATM.
Previous | Table of Contents | Next |