ATM in Europe: The User Handbook

2 Basic ATM technical characteristics*

2.1 ATM - the background
There has always been a need for more and cheaper bandwidth, but never more so than today. This need has been accelerated by the availability of high-performance and relatively low-cost PCs and workstations and business software applications. Indeed there is a direct correlation between MIPS and bandwidth required to run applications. This distributed approach to computing demands higher bandwidth at lower cost, available on both wide area and metropolitan networks (WANs and MANs). Public Network Operators (PNOs) and carriers are faced with the challenge of providing easy to manage broadband services when required (on demand) and at an affordable price. PNOs saw fibre optic cable as the answer to this dilemma with the potential of almost unlimited bandwidth. Leading PNOs have been investing heavily in the installation of fibre, replacing copper as the basic communications medium of the future. ITU-TS (formerly CCITT) and other standards groups started work in the study period 1984-1988 to establish a series of recommendations for the transmission, switching, signalling and control techniques required to realise an intelligent fibre-based
* Contents are based on the following documents: [9] Introduction to ATM, ATMF white paper; [10] ATM Service Categories: the Benefits to the User, ATMF EMAC white paper; [11] Broadband ISDN Protocol Reference Model, ATMF EMAC contribution, Egidio Perretti, Telecom Italia, May 1996.

network. Decisions were made during this period to base the Broadband Integrated Services Digital Network, or B-ISDN, on SDH and ATM. SDH (Synchronous Digital Hierarchy) describes the optical standards for transmission of data at rates up to 2.4 Gbit/s with the expectation to reach 10 Gbit/s in the future. SDH describes how lower-speed data can be carried or mapped into the optical network and multiplexed. An essential element is to ensure that equipment and services from different vendors/service providers can fully interoperate. Because SDH describes the transmission of data it can operate with nearly all emerging switching technologies. In particular ITU-TS specifications detail how ATM cells are mapped into SDH. Maximising the potential of optical broadband transmission generates the need for an intelligent switching fabric with the ability to switch very quickly all forms of traffic (voice, data, image, multimedia) whilst maximising the use of bandwidth. Ideally, bandwidth should be shared between applications and allocated on demand. ATM has been selected by the worlds PNOs as the switching technology which will ultimately satisfy these stringent criteria. In July 1996, a major milestone in the development of ATM technology was

reached when The ATM Forum announced agreement on the Anchorage Accord. The Anchorage Accord contains the Foundation Specs for mission-critical ATM infrastructure, as well as Expanded Feature Specs for migration to ATM multi-service networks. In the future, Foundation Specs will be only revised to correct problems, align with ITU-T recommendations, and add cohesiveness/ parity between specifications. Any new specifications will be backwards compatible to Anchorage Accord specifications. (For information on how to order a CD-ROM containing the Anchorage Accord specs, see Appendix 1.)

MANAGEMENT PLANE CONTROL PLANE Higher Layer Protocols USER PLANE Layer Management Functions Higher Layer Protocols Plane Management Functions

ATM ADAPTATION LAYER

ATM LAYER (**)

PHYSICAL LAYER

Transm. Convergence sublayer Physical medium sublayer

(**) Common for all services

Fig. 2.1 B-ISDN protocol reference model. Source: ITU-T; I.371

2.2

B-ISDN protocol reference model

The B-ISDN protocol reference model defined by ITU-T is shown in figure 2.1; it is composed of a user plane, a control plane and a management plane, each of them responsible for associated functions. In particular: The user plane, with its layered structure, provides for user information flow transfer, along with associated controls ranging from flow control to error recovery, etc.; The control plane has a layered structure and performs the call control and connection control functions; it deals with the signalling necessar to set up, supervise and release calls and connections; The management plane provides two different types of functions: - plane management functions, not layered, that are related to a system

as a whole an provide co-ordination between all the planes and - layer management functions, that are related to resources and parameters residing in its protocol entities; layer management handles the operation and maintenance (OAM) information flows specific to the layer concerned. In terms of ATM protocol stack, above the Physical Layer there is the ATM Layer that provides call transfer for all services and the ATM Adaptation Layer (AAL) providing service-dependent functions to the layers above (indicated as higher layers). The layer above AAL in the control plane provides call control and connection control; the management plane provides network supervision functions. The functions of each layer are detailed in figure 2.2, which also shows sublayers: Convergence Sublayer (CS) and Segmentation and Reassembly Sublayer (SAR) for the AAL, Transmission Convergence (TC) and Physical Medium (PM) for the PHY.

ATM in Europe: The User Handbook

Higher layer functions Convergence Segmentation & Reassembly Layer Management Generic flow control Cell header generation/extraction Cell VIP/VCI translation Cell multiplex and demultiplex Cell rate decoupling HEC head.sequ. generation/verification Cell delineation Transmission frame adaptation Transmission frame generation/recovery Bit timing Physical medium
Fig. 2.2 Functions of the B-ISDN in relation to the Protocol Reference Model. Source: ITU-T; I.371

Higher CS SAR

layers AAL

ATM

TC

Physical layer

PM

2.3

ATM cells and the Adaptation Layer concept

networks task is only to route the cell from one point to another, depending on its header information. It should be noted that up to four bytes may be used by the adaptation process itself with some AAL types, leaving 44 bytes for payload information. Several adaptation layers have already been standardised. These are: Type 1: Constant Bit Rate (CBR) services. AAL1 handles traffic where there is a strong timing relation between the source and the destination. Examples include PCM-encoded voice traffic, constant bit rate video and the emulation of public network circuits (e.g. the transport for E1 links). Type 2: Variable Bit Rate (VBR) timingsensitive services. AAL2 handles traffic where a strong timing relation between the source and the destination is

ATM is a packet technology that directs traffic using a label contained in the packets header. Unlike other packet technologies, such as X.25 or frame relay, ATM uses short, fixed-length packets called cells. Each cell is 53 bytes long: 48 bytes for the information field and 5 bytes for the preceding header (see fig. 2.3). The header field contains information about the virtual channel (VCI: Virtual Channel Identifier) and virtual path (VPI: Virtual Path Identifier) in use, payload type (PT) and cell loss priority (CLP). Inserting payload data into the 48-byte information field of the ATM cell is accomplished by the ATM Adaptation Layer (AAL). The AAL is what gives ATM the flexibility to carry entirely different types of services within the same format. It is important to understand that the AAL is not a network process but instead is performed by the network terminating equipment. Thus the

1 octet VPI 5-byte header VPI VCI VCI HEC PT CLP VCI

48-byte payload

INFORMATION FIELD

Fig. 2.3 ATM cell format. Source: ITU-T; I.371

required, but the bit rate may vary. Examples include variable bit rate voice and compressed, for instance MPEGcoded, video. Type 3/4: Connection-oriented and connection-less VBR data transfer. AAL3/4 is a fairly complex layer that can handle VBR (i.e. bursty) data both with and without pre-establishing an ATM link. Examples for the connectionoriented type include large file transfers like CAD files or data back-up. The connectionless type is intended for short, highly bursty transfers as might be generated by LANs. Type 5: Simple and Efficient Adaptation Layer (SEAL). AAL5 may be looked upon as a simplified version of AAL3/4 that is designed to meet the requirements of local, high-speed LAN implementations. AAL5 is intended for connectionless or connection-oriented VBR services.

2.4

ATM networks

A highly simplified example for the structure of an ATM network is shown in Figure 2.4. It is important to understand that the various UNI and NNI connections could be carried via different physical media, such as the existing Plesiochronous Digital Hierarchy (PDH) layers or the new Synchronous Digital Hierarchy (SDH). Several standards have been defined on how to interface the physical layers and work is continuing to specify additional physical layers to be used to transport ATM cells. This structure can be the same for local area networks (LANs), metropolitan area networks (MANs), and wide area networks (WANs). However, there may be numerous media, physical layers, protocols and services, all within the same infrastructure. This also means that there will be a need for connection-oriented and connectionless

ATM in Europe: The User Handbook

traffic to run in parallel. ATM is designed to support all these requirements. While it is beyond the scope of this handbook to give a detailed description of how ATM cells are transported over the network, the concept of virtual circuits, which are known as Virtual Channel Connections (VCCs), can be described in the following way: A VCC is set up between any source and any destination in the ATM network, regardless of the way it is being routed across the network. Fundamentally, ATM is a connection-oriented technology. The way the network sets up the connection is therefore by signalling, i.e. by transmitting a set-up request which passes across the network to the destination. If the destination agrees to form a connection, the VCC is set up between the two end-systems. A mapping is

defined between the Virtual Channel Identifiers (VCIs) / Virtual Path Identifiers (VPIs) of both UNIs, and between the appropriate input link and the corresponding output link of all intermediate switches. A VCC is a connection between two communicating ATM end-entities. It may consist of a concatenation of several ATM VC links. All communication proceeds along this same VCC which preserves cell sequence and provides a certain quality of service. Note that the Virtual Channel Identifier (VCI) in the ATM cell header is assigned per network entity-to-entity link, i.e. it may change across the network within the same VCC (see Figure 2.5). A Virtual Path (VP) groups VCs carried between two ATM entities and may also involve many ATM VP links. The VCs associated with a VP are globally switched

CUSTOMER PREMISES EQUIPMENT (CPE) UNI UNI SWITCH NNI SWITCH

UNI NNI ATM SWITCH

NNI: Network-Node Interface UNI: User-Network Interface

UNI NNI ATM

UNI UNI ATM

Fig. 2.4 ATM Network. Source: Telecom Italia

VCI a

VPI X
VCI b

Transmission Payload
VCI a

VPI Y
VCI b

Fig. 2.5 Virtual channel and virtual paths. Source: ITU-T

without unbundling or processing the individual VC in any way or changing their VCI numbers. Thus the cell sequence of each VC is still preserved and the quality of service of the VP depends on that of its most demanding VC. As the cell address mechanism uses both the VCI and the VPI, different VPs may also use the same VCI without conflict. A cell may also not be associated with any VP. In this case it would have a null VPI and only a unique VCI. By means of VCs and VPs, virtual circuits can be set up either permanently (by using so-called Permanent Virtual Channels, PVCs) or on demand (Switched Virtual Channels, SVCs). It is likely that VPs will be used mostly between switches (i.e. across NNIs) to carry across large numbers of virtual circuits. In any case, all the ATM switch has to do is to identify, on the basis of the cells VPI, VCI or both, which output a received cell needs to be routed to and what the new VPI/ VCI on this output link is. The operation of an ATM network is

therefore very simple and inherently can scale to very high speeds.

2.5 ATM services

The introduction of new ATM service categories increased the benefits of ATM, making the technology suitable for a virtually unlimited range of applications. An ATM network can provide Virtual Path (VP) or Virtual Channel (VC) Connections with different levels of service. The concept of negotiating the behaviour expected from the ATM layer in terms of traffic and performance for each connection allows users to optimise network capabilities to suit the applications requirements. The first ATM implementations offered limited options. A typical network behaviour, common to most of the first generation ATM networks, is to reserve a fixed amount of bandwidth for each connection for the duration of the call on the basis of the

ATM in Europe: The User Handbook

maximum emission rate of the source (i.e. the peak cell rate, PCR) and to provide a single level of quality of service. The ATM service categories represent new service building-blocks that make it possible for users to select specific combinations of traffic and performance parameters. ATM is a multiservice technology. Actually, most of the requirements that are specific to a given application may be resolved at the edges of an ATM network by choosing an appropriate ATM Adaptation Layer (AAL). However, in accordance with the standards definitions the ATM-layer behaviour should not rely on the AAL protocols since these are service specific (and are in many cases supported by the user terminal, i.e. outside the core network visibility), nor on higher layer protocols which are application specific. Given the presence of a heterogeneous traffic mix and the need to adequately control the allocation of network resources for each traffic component, a much greater degree of flexibility, fairness and utilisation of the network can be achieved by providing a selectable set of capabilities within the ATM-layer itself. The Service Categories have been defined with this goal in mind. Both users and network operators can benefit from the availability of a selectable set of ATM-layer services. These services are, in effect, the tools which will allow the promise of ATM to be fully met: Customer perspective: ATM customers (e.g. end-users, IT and telecommunications managers) aim to save on network usage costs, provided that their substantial efficiency and quality requirements are matched. Require-

ments vary in nature depending on what application (e.g. data, voice, video, multimedia) is running. As a matter of fact users that produce variable traffic patterns would like to be able to get bandwidth just when actually needed and, in case of elastic sources, to have fast access to as much available bandwidth as possible, achieving a satisfactory compromise between performance and cost. Network and service operators perspective: All types of operators investing in ATM infrastructures and services aim to achieve maximum use of the deployed resources, avoiding congestion while being able to share network resources among a large number of customers and fulfilling the different user needs in a cost-effective way. This allows for appropriate tariffing strategies to be deployed. The ability to offer a range of network services, with selectable cost/performance levels, is a key issue for network operators, particularly in a competitive market.

2.6 ATM service categories and applications

Constant Bit Rate (CBR) The CBR service category is used by connections that require a static amount of bandwidth, characterised by a Peak Cell Rate (PCR) value that is continuously available during the connection lifetime. The source may emit cells at or below the PCR at any time and for any duration (or may be silent). This category is intended for real-time applications, i.e. those requiring tightly constrained Cell Transfer Delay (CTD) and Cell Delay Variation (CDV), but is not

restricted to these applications. It would be appropriate for voice and video applications as well as for Circuit Emulation Services (CES). The basic commitment made by the network is that once the connection is established the negotiated QoS is assured to all cells conforming to the relevant conformance tests. Cells which are delayed beyond the value specified by Cell Transfer Delay (CTD) are assumed to be of significantly less value to the application. For instance a cell received with excessive delay may not be of use to decoders for reconstructing an isochronous flow and should be considered as lost. Typical Applications for CBR Any data/text/image transfer application which contains sufficiently smooth traffic or for which the end-systems response time requirements justify occupying a fully reserved CBR channel. Examples are: - Video-conferencing - Interactive Audio (e.g. telephony) - Audio/Video Distribution (e.g. television, distance learning, pay-per-view) - Audio/Video Retrieval (e.g. video-on demand, audio library) For telephony and voiceband services over ATM, e.g. 64 kbit/s N-ISDN-compatible services, the access solution based on AAL1 requires CBR support for taking advantage of delay and variance bounds. In the multimedia area, a near-term solution for residential services foresees VoD based on MPEG2 (Transport Stream, CBR mode) over AAL5 with transportation being provided by the ATMlayer with CBR service. Real-Time Variable Bit Rate (rt-VBR) The real-time VBR service category is

intended for real-time applications, (i.e. those requiring tightly constrained delay and delay variation), as would be appropriate for voice and video applications. Sources are expected to transmit at a rate that varies with time. Alternatively, the source can be described as bursty. Traffic parameters are Peak Cell Rate (PCR), Sustainable Cell Rate (SCR) and Maximum Burst Size (MBS). Cells that are delayed beyond the value specified by CTD are assumed to be of significantly less value to the application. Real-time VBR service may support statistical multiplexing of real-time sources. Non-Real-Time (nrt-VBR) The non-real time VBR service category is intended for non-real time applications that have bursty traffic characteristics and can be characterised in terms of a PCR, SCR and MBS. For those cells that are transferred within the traffic contract the application expects a low Cell Loss Ratio (CLR). For all cells it expects a bound on the Cell Transfer Delay (CTD). Non-real time VBR service may support statistical multiplexing of connections. Typical applications for VBR VBR is suitable for any application for which the end-system can benefit from statistical multiplexing by sending information at a variable rate and can tolerate or recover a potentially small random loss ratio. This is the case for any constant bit rate source for which variable rate transmission allows more efficient use of network resources without performance impairment. Real-time VBR, in particular, can be used by native ATM voice with bandwidth compression and silence suppression. For some classes of multimedia communications real-time VBR may be very appropriate. Non-real time VBR can be used for data transfer, e.g. for response-time

10

ATM in Europe: The User Handbook

critical transaction processing applications (e.g., airline reservations, banking transactions, process monitoring) and frame relay interworking. Available Bit Rate (ABR) ABR is an ATM layer service category for which the limiting ATM layer transfer characteristics provided by the network may change subsequent to connection establishment. A flow control mechanism is specified which supports several types of feedback to control the source rate in response to changing ATM layer transfer characteristics. Many sources (applications) have the ability to reduce or increase their information rate if the network requires them to do so. It is expected that an end-system that adapts its traffic in accordance with the feedback will experience a low Cell Loss Ratio (CLR) and obtain a fair share of the available bandwidth according to a network specific allocation policy. Cell Delay Variation (CDV) is not controlled in this service, although admitted cells are not delayed unnecessarily. ABR service is not intended to support real-time applications. On the establishment of an ABR connection the end-system shall specify to the network both a maximum required bandwidth and a minimum usable bandwidth. These shall be designated as Peak Cell Rate (PCR) and the Minimum Cell Rate (MCR) respectively. The MCR may be specified as zero. The bandwidth available from the network may vary, but shall not become less than MCR. A source, destination and network node behaviour is specified along with details of a rate-based flow control mechanism. Typical applications for ABR Any non-time critical application running over an end-system capable of varying its emission rate can exploit the ABR service. This category provides an economical support

to those applications that show vague requirements for throughput and delay and require a low cell loss ratio (CLR). Examples include LAN interconnection / internetworking services, which are driving the business service market for ATM. These are typically run over router-based protocol stacks like TCP/IP, which can easily vary their emission rate as required by the ABR rate control policy. The support through ABR is likely to result in an increased end-to-end performance (goodput). Another application environment suitable for ABR is LAN Emulation. Other application examples are critical data transfer (e.g. defence information, banking services), supercomputer applications, and data communications, such as remote procedure call, distributed file services and computer process swapping/paging. Unspecified Bit Rate (UBR) The Unspecified Bit Rate (UBR) service category is intended for non-real-time applications, which do not require tightly constrained delay and delay variation. Examples of such applications are traditional computer communications applications, such as file transfer and e-mail. UBR sources are expected by nature to transmit non-continuous bursts of cells, otherwise a traffic shaping algorithm is required. UBR service supports a high degree of statistical multiplexing among sources. UBR service does not specify trafficrelated service guarantees. Specifically, UBR does not include the notion of a perconnection negotiated bandwidth. There may not be any numerical commitments made as to the cell loss ratio experienced by a UBR connection or as to the cell transfer delay experienced by cells on the connection. Typical applications for UBR UBR can provide a suitable solution for less demanding applications. Most data

11

APPLICATION AREA Critical Data LAN interconnection LAN emulation Data Transport/interworking (IP-FR-SMDS) Circuit Emulation-PABX POTS/ISDN -video-conference Compressed Audio Video Distribution Interactive Multimedia

CBR $$ $

rt-VBR nrt-VRB $ $ $$$ $$

ABR $ $$$

UBR N/S $$

$ $$$ $$$ $ $$$ $$$

$ $$

$$ N/S

$$$ N/S N/S

$$ N/S N/S $ N/S $

$$$ $$ $$$

$$ $ $$

$$ N/S $$

Score to indicate the advantage: Optimum: $$$ Good: $$

Fair: $ N/S: Not suitable Not quoted: Presently considered not applicable with advantage (might be in the future)

Table 2.1

Application areas for ATM service categories. Source: ATMF white paper: ATM Service Categories: The Benefits to the User.

applications, e.g. file transfer submitted in the background of a workstation with minimal service requirements, are very tolerant to delay and cell loss (store and forward networks are in fact widely used for these applications). Examples may include: - Text/Data/Image Transfer, Messaging, Distribution, Retrieval - Remote Terminal (e.g. telecommuting) The above services can take advantage of any spare bandwidth and will profit from the resultant reduced tariffs (cheap services).

Applications summary Table 2.1 is an attempt to sum up the indications outlined in this section. The association and the score assignment are based on a subjective perception (shared among a number of people working on this subject). However this is not intended to create a restrictive correspondence between the identified application areas and ATM-layer services and should not be taken restrictively.

12