This lecture is divided into hyperlinked sections
Communication
Switching Techniques
Circuit Switching
Multirate Circuit
Switching
Packet Switching
Circuit-Switched
shortcomings
The
Improvements of Frame Relay
Frame Relay and
Cell Relay
Frame Relay technology
Frame
Mode Bearer Service And Protocol
Why will
ATM provide a solution?
Asynchronous
Transfer Mode - key features
Cell Relay
Cells and Switching
Asynchronous TDM
Virtual Circuit
Connections
Virtual
Channels and Virtual Paths
ATM Forum
Conclusion
Figure 7.1 shows a spectrum of switching techniques available to transport information across a network. The two extreme ends of the spectrum represent the two traditional switching techniques: circuit switching and packet switching. The remaining techniques are of more recent vintage.
In general, the techniques toward the left end of the line provide transmission with little or no variability and with minimal processing demands on attached stations, while techniques toward the right end provide increased flexibility to handle varying bit rates and unpredictable traffic at the expense of increasing processing complexity.
Fig 7.1 Spectrum of switching techniques
Circuit switching is the traditional technique used for telephone networks and is the basic technology for narrowband ISDN. The essence of circuit switching is the establishment of a fixed capacity circuit for the complete duration of a connection. To accommodate different data rate requirements and/ or multiple users at an endpoint, a number of circuits can be multiplexed in a fixed, synchronous time division multiplexing structure.
· When the standards work on broadband ISDN (BISDN) began in the mid 1980s, it was generally assumed by most participants that some form of synchronous time division multiplexing (TDM) technique would be used, as is the case with the basic and primary rate access methods for ISDN. Under this approach, the interface structure that was proposed was
j x H4 + k x H2 + l x H1 + m x H0 + n x B + D
where D, B, H0, and H1 (H11 and H12) are narrowband ISDN channels and H2 and H4 are new BISDN fixed rate channels. H2 would be in the range of 30 to 45 Mbps and H4 in the range of 120 to 140 Mbps.
This allows a customer to build a data channel to their own requirements by choosing suitable values for j, k, l, m, n.
Table 7. 1 Comparison of ISDN rates
Although the synchronous TDM approach is a natural extension of narrowband ISDN, it does not provide the best model for BISDN. There are two basic disadvantages of the synchronous approach.
· It does not provide a flexible interface for
meeting a variety of needs. At the high data rates offered by BISDN, there
could be a wide variety of applications, and many different data rates,
that need to be switched. One or two fixed rate channel types do not provide
a structure that can easily accommodate this requirement. Many data (as
opposed to voice or video) applications are bursty in nature and can more
efficiently be handled with some sort of packet switching approach.
· The second disadvantage of the synchronous approach
for high-speed transmission is that the use of multiple high data rates
(e.g. a number of H2 and H4 channels) complicates the switching system.
It would require switches that can handle data streams of multiple high
data rates. This is in contrast to narrowband ISDN, which has just the
64kbps data stream to switch.
An enhancement of the traditional synchronous TDM approach has been investigated that seeks to provide more flexibility. This technique is known as multirate circuit switching. The transmission technique for multirate circuit switching is the same as that for pure circuit switching. The enhancement provided by multirate circuit switching is the ability to construct a connection consisting of multiple, synchronised circuits. For example, the video codecs developed for narrowband ISDN, and standardised by Recommendation H.261, can operate at bit rates that are multiples of the 64kbps B-channel circuit rate.
The multirate circuit switching approach does provide some increased level of flexibility at the cost of increased switching and access complexity. This will support a variety of applications that require different sustained data rates. However, it still does not provide an efficient means of supporting bursty traffic.
The traditional approach to packet switching is X.25. Several key features of the X.25 approach are:
· Call control packets, used for setting up and
clearing virtual circuits, are carried on the same channel and same virtual
circuit as data packets. In effect, inband signalling is used
· Virtual circuits can be set up at layer 3
· Both layer 2 and layer 3 include flow-control
and error-control mechanisms.
This approach results in considerable network overhead.
Figure 7.2 indicates the flow of data link frames required for the transmission
of a single data packet from
source-system to destination-system and the return of
an acknowledgement packet. At each hop through the network, the data link
control protocol involves the exchange of a data frame and an acknowledgement
frame.
It can be seen that the data packet and the acknowledgement packet generate data link acknowledgement frames during their progress through the network.
Figure 7.2 Packet
switched network
Furthermore, at each intermediate node, state tables must be maintained for each virtual circuit to deal with the call management and flow control/error control aspects of the X.25 protocol.
All of this overhead may be justified when there is a significant probability of error on any of the links in the network. This approach may not be the most appropriate for ISDN. On the one hand, ISDN employs reliable digital transmission technology over high quality, reliable transmission links, many of which are optical fibre. On the other hand, with ISDN, high data rates can be achieved, especially with the use of H channels. In this environment, the overhead of X.25 is not only unnecessary, but degrades the effective utilisation of the high data rates available with ISDN.
Thus the X.25 approach results in too much network chatter when the line that the signals are being carried over is reliable, because of the considerable overhead in the protocol. This would mean that it is difficult to implement an X.25 based network to support applications that require high data rates.
Secondly, because of the variable delay in delivering packets, X.25 is not suitable for isochronous applications, such as voice, that require a constant bit rate. This does not mean that voice cannot be carried, but the variable delay means that conditions are not ideal.
Packet switching was developed at a time when digital long distance transmission facilities exhibited a relatively high error rate compared to today's facilities. As a result, there is a considerable amount of overhead built into packet switching schemes to compensate for errors. The overhead includes additional bits added to each packet to enhance redundancy and additional processing at the end stations and the intermediate network nodes to detect and recover from errors.
With modern, high-speed telecommunications systems, this overhead is unnecessary and counterproductive. It is unnecessary because the rate of errors (BER) has been dramatically lowered, and any remaining errors can easily be caught by logic in the end systems that operate above the level of the packet switching logic. It is counterproductive because the overhead involved soaks up a significant fraction of the high capacity provided by the network.
The long haul circuit-switched telecommunications network was originally designed to handle voice traffic, and the majority of traffic on these networks continues to be voice. One of the key characteristics of circuit switched networks is that resources within the network are dedicated to a particular call. For voice connections, the resulting circuit will enjoy a high percentage of utilisation since most of the time, one party or the other is talking.
However, as the circuit switched network began to be used increasingly for data connections, two shortcomings became apparent.
· In a typical terminal-to-host data connection, much of the time the line is idle. Thus, with data connections, a circuit switched approach is inefficient.
· In a circuit-switched network, the connection provides for transmission at constant data rate. Thus each of the two devices that are connected must transmit and receive at the same data rate as the other to efficiently use the bandwidth of the circuit-switched connection. This limits the usefulness of the network in interconnecting a variety of host computers and terminals.
Frame Relay attempts to improve matters (read more)
Frame relaying is designed to eliminate as much as possible of the overhead of X.25. The key differences of frame relaying from a conventional X.25 packet switching service are:
· Call control signalling is carried on a separate
logical connection from user data. Thus intermediate nodes need not maintain
state tables or process messages relating to call control on an individual
per-connection basis.
· Multiplexing and switching of logical connections
take place at layer 2 instead of layer 3, eliminating one entire layer
of processing.
· There is no hop-by-hop flow control and error
control. End-to-end flow control and error control are the responsibility
of a higher layer, if they are employed at all.
Figure 7.3 indicates the operation of frame relay, in which a single user data frame is sent from source to destination, and an acknowledgement, generated at a higher layer, is carried back in a frame.
Fig 7.3 Frame relay: source sending, destination responding
Let us consider the advantages and disadvantages of this approach. The principal potential disadvantage of frame relaying, compared to X.25, is that we have lost the ability to do link-by-link flow and error control. Although frame relay does not provide end-to-end flow and error control, this is easily provided at a higher layer. In X.25, multiple virtual circuits are carried on a single physical link, and LAPB is available at the link level for providing reliable transmission from the source to the packet switching network and from the packet switching network to the destination. In addition, at each hop through the network, the link control protocol can be used for reliability. With the use of frame relaying, this hop-by-hop link control is lost. However, with the increasing reliability of transmission and switching facilities, this is not a major disadvantage.
The advantage of frame relaying is that we have streamlined the communications process. The protocol functionality required at the user network interface is reduced, as is the internal network processing. As a result, lower delay and higher throughput can be expected. Preliminary results indicate a reduction in frame processing time of an order of magnitude, and the CCITT recommendation (I.233) indicates that frame relay is to be used at access speeds up to 2 Mbps. Thus, we can expect to see frame relaying supplant X.25 as ISDN matures.
The ANSI standard T1.606 lists some examples of applications that would benefit from the frame relay service used over a high speed H channel:
Block interactive data applications: An example of a block interactive application would be high-resolution graphics (e.g. high resolution videotex, CAD/CAM). The pertinent characteristics of this type of application are low delays and high throughput.
File transfer: The file transfer application is intended to cater to large file transfer requirements. Transit delay is not as critical for this application as for real time applications.
Character-interactive traffic: An example is text editing characterised by short frames, low delays and low throughput.
Perhaps the most important development in communications networking over the past few years has been the introduction of frame relay and cell relay capabilities into various networking and communications products. Frame relay represents the first major improvement in packet switching technology in over twenty years. Cell relay, which follows the same evolutionary path as frame relay, is seen by many in the industry to be equivalent to networking nirvana. Indeed, it is hard to overstate the enthusiasm generated by these two technologies.
Frame relay and cell relay are outgrowths of the work on ISDN and BISDN.
To take advantage of the high data rates and low error rates of contemporary networking facilities, frame relay was developed. Whereas the original packet switching networks were designed with a data rate to the end user of about 64 kbps, frame relay networks are designed to operate at user data rates of up to 2 Mbps. The key to achieving these high data rates is to strip out most of the overhead involved with error control.
Frame Mode Bearer Service And Protocol
CCITT has published two recommendations for frame relay:
· 233: ISDN Frame Mode Bearer Services (1991).
· 370: Congestion Management for the Frame Relaying
Bearer Service. (1991).
The work on frame relay is more developed in the United States, where ANSI has issued three standards:
· ANSI T1.606: Architectural Framework and Service
Description for Frame Relaying Bearer Service (1990).
· Draft ANSI T1.617: Signalling Specification
for Frame Relay Bearer Service (1991).
· ANSI T1.618: Core Aspects of Frame Protocol
for Use with Frame Relay Bearer Service (1991).
It is anticipated that the final CCITT recommendations will be closely aligned with the current ANSI standards.
Why will ATM provide a solution?
The trend of applications to produce larger and more complex output has led to a demand for a data transfer system that can adapt to the various types of network traffic generated today. ATM is seen by many as an ideal solution.
Users desire some sort of simplification so that many different types of traffic and services can be carried across networks speedily. The technology exists now with high speed ICs, processors and memory becoming ever cheaper. Cabling technology has improved greatly with optical fibre, UTP and ScTP technologies constantly advancing.
Interest is growing amongst users, vendors, manufacturers and service providers who can see the benefit of ATM as a unifying technology.
Asynchronous Transfer Mode - Key Features
· ATM is a new communication and networking technology
that builds on the work of ISDN and uses the latest technologies.
· ATM is the switching of fixed size cells through
a network via virtual connections.
· ATM uses asynchronous multiplexing which is
a kind of statistical multiplexing.
· ATM promises scalability with the same solution
for LANs, MANs and WANs.
· ATM also promises universal data transport for
data, images, voice, video and multimedia. It also can handle asynchronous
and isochronous traffic.
Cell Relay, also known as asynchronous transfer mode (ATM), is in a sense a culmination of all of the developments in circuit switching and packet switching over the past twenty years. One useful way to view cell relay is as an evolution from frame relay. Both frame relay and ATM take advantage of the reliability and fidelity of modern digital facilities to provide faster packet switching than X.25.
Like frame relay and X.25, cell relay allows multiple logical connections to be multiplexed over a single physical interface. This means one connection and over this can run many different types of network traffic.
As with frame relay, cell relay has no link-by-link error control or flow control. Cell relay provides minimum overhead for error control, depending on the inherent reliability of the transmission system and on higher layers of logic to catch and correct remaining errors.
The most obvious difference between cell relay and frame relay is that frame relay uses variable length packets, called frames, and cell relay uses fixed length packets, called cells. By using a fixed packet length, the processing overhead is reduced even further for cell relay compared to frame relay. The result is that cell relay is designed to work in the range of 100s and 1000s of Mbps, compared to the 2 Mbps of frame relay.
Table 7.2 Comparison of X.25, Frame relay and Cell relay
Another way to view cell relay is as an evolution from multirate circuit switching. With multirate circuit switching, only fixed data rate channels are available to the end system.
Cell relay allows the definition of virtual channels with data rates that are dynamically defined at the time that the virtual channel is created. By using small, fixed size cells, cell relay is so efficient that it can offer a constant data-rate channel even though it is using a packet-switching technique. Thus cell relay extends multirate circuit switching to allow multiple channels with the data-rate of each channel dynamically set on demand.
All ATM cells are 53 bytes long, 48 for data and 5 for the cell header. Service adapters convert the data stream to and from the ATM environment. The adapters exist to deal with data, voice video etc. Because ATM can carry different types of traffic, some sort of priority scheme is needed to help cope with delivery of the cell streams.
From the adapters the data are fed into asynchronous multiplexers to combine the cells for high data-rate transport. All the tributary cell streams may be unrelated in timing and data rate.
A connection must be established before any data transfer may take place. ATM is always connection oriented so that once the connection path has been established between both communicating entities, all cells will follow the same path. These connections are known as virtual connections. This cuts out the latency at each hop through the network experienced by the datagram approach to communication.
Fig 7.4 ATM Cells multiplexed onto carrier
All cells are 53 bytes long, this value being a compromise between the computing and telecomms communities. By having all cells the same size allows for hardware design to be optimised so that custom hardware can be designed to deal with fixed size cells. The small size is better for isochronous and asynchronous traffic to mix together.
The header is 5 bytes long and carries the transport information. The payload is 48 bytes and can carry user data, signalling and line management information. The cell rate will vary depending on the data rate of the cable on which they are being carried.
Table 7.3 Cell rates compared
The small size of the cells allows support for multimedia applications so that voice, video, images and data can be carried simultaneously. This means that cells can arrive frequently and predictably.
The processing required for routing/ switching is carried out exclusively in hardware, not software and so is faster. Both service access and bandwidth can be guaranteed.
If we compare this against other technologies that use large frames and packets that have a large payload to overhead ratio (Ethernet has 1500 bytes payload and 18 bytes of overhead) the goal is throughput efficiency when bit rates are low. The network latency can be seen to increase for large frames/ packets with processing, routing and forwarding being time consuming. Jitter will accumulate and latency is widely variable with traffic.
ATM makes use of Asynchronous Time Division Multiplexing ATDM to insert cells onto a transmission line. The tributary signals that are low speed when compared with the carrier are formatted into streams of cells and may be either streamed or burst. Tributary cells are interleaved on an as needed basis, obeying the priority assigned to each stream of cells.
Fig 7.5 Asynchronous multiplexing of unrelated data
The cells arrive at the asynchronous multiplexer and are stored in a buffer or buffers and are then inserted into a multiplex slot according to the priority of the cell streams. The priority rules are arranged such that real-time data traffic, i.e. isochronous has a higher priority for delivery than bursty file transfers.
ATDM has a higher channel utilisation than TDM with TDM tributary channels always occupying the same position in the slot in successive frames. ATDM cells occupy the slots on an as-needed basis.
Because ATM is a connection oriented technology, cells are transported along virtual circuit connections VCCs and data is only able to flow after the VCC has been made. The VCC can be:
· Point-to-point bi-directional (Full duplex) connections
i.e. unicast
· One-to-many i.e. multicast
· One-to-any i.e. anycast
Bi-directional VCCs can have symmetric or asymmetric data rates depending on the requirements of the connection.
The VCCs may be either Permanent Virtual Circuits PVCs or Switched Virtual Circuits SVCs. PVCs are long duration connections that are assigned and released by the network administrator and use proprietary GUI software from the equipment vendor. SVCs on the other hand are dynamically assigned by the network and use proprietary route-calculation software. Each SVC is released or erased after use. Bandwidth can be supplied on demand.
Virtual Channels and Virtual Paths
VCCs are identified by two numbers, Virtual Channel Identifier VCI and Virtual Path Identifier. A VCI refers to a virtual channel VC and this is a source to destination connection. VCs have a specified Quality of Service QoS parameter. The parameter encompasses such items as data rate, isochronous, error rate and delay tolerance. A virtual path VP is a bundle of VCs and is done to increase transport efficiency by creating a hierarchy to help switching a whole set of VCs.
Figure 7.6 Virtual Channels and Virtual Paths
The ATM forum was created in 1991 to promote co-operation among manufacturers and has around 1000 members world-wide comprising of companies from all market segments. More information in PDF format may be found at http://www.atmforum.com.
The objectives of the forum are to bring ATM products to the marketplace as soon as possible and to ensure early interoperability between products from different vendors. It also seeks to bring about early implementation at LAN, MAN and WAN levels and to educate users of the benefits of ATM technology.
The forum meets monthly and releases specifications i.e. implementation agreements and influences the development of standards by the ITU-T.
The switching techniques of circuit switching, multirate
circuit switching and packet switching do not entirely meet all the criteria
requested by various applications.
The overhead associated with X.25 makes it unsuitable
for high speed communications by allowing for error checking at each hop
of its journey.
ATM and saw that uses fixed length cells and that this
allows for optimised hardware with speed gains.
Virtual connections are established between pairs of
end points and this cuts the latency at each hop that occurs with the X.25
and datagram approaches.
ATDM is used to transport these cells and that there
are priorities assigned to the bit streams according to the needs of the
application.
Virtual paths exists between two end entities and may
contain many virtual channels.
© M Clements 2000