Broadband Primer – Computer Data Transfer
Fast, reliable communications is becoming increasingly important in today’s business world. Gathering information, analysing it, and making that deadline, can be a key competitive advantage to your organisation, all possible in 2017 with computer data transfer. With today’s global markets, this information must be accessible over large geographic areas. What is required is a communications network that is capable of meeting these needs. Broadband-ISDN is such a network.
This primer describes the type of services that will run over such a Broadband-ISDN network. Asynchronous Transfer Mode (ATM), the technology required to realise such a network, is also explained.
This primer is an introduction to the subject from a non technical standpoint. It assumes that you are familiar with present day telecommunications techniques, at least at a conceptual level.
The primer has a number of sections. The Broadband services section describes the type of services that will become available. Then the underlying Broadband-ISDN technology, Asynchronous Transfer Mode is explained in some detail. Next ATM deployment topologies are investigated, both in the local and wide area. A conclusion provides a summary of the preceding sections.
Broadband-ISDN Services For Transfering Computer Data
With the deployment of Broadband-ISDN for fast reliable computer data transfer, it will be possible for the first time to have a network capable of supporting both low speed and very high speed services; these services include voice, data, video and image.These services have a variety of bandwidth requirements ranging from 64kbit/s to 600Mbit/s. They can be constant bit rate (CBR) services or variable bit rate (VBR) services, depending upon application. Services such as voice, require a pre-determined transmission rate to retain their service quality and are known as isochronous services. Other services such as computer data require data integrity and can be “bursty” in nature. Broadband-ISDN networks will handle all these services.
A major application of Broadband-ISDN networks will be high speed data communication. Present computer local area networking applications are accomplished over legacy LAN systems (Ethernet and Token Ring) which have limited reach and speed. These applications can be extended in performance and geography through a Broadband-ISDN network. Backwards capability of applications is a key issue.
Another application area of Broadband-ISDN networks is video. Video telephony permits image as well as voice to be exchanged between participants. Videoconferencing extends the video telephony concept; multiple participants in different locations will be able to communicate with each other in a conference call using video.
Present broadcast networks, such as cable TV, are passive as far as the user is concerned. The Broadband-ISDN networks will permit interactive services. Video-on-Demand (VoD) will allow the user to select a movie of their choice and play it when they like rather than having to rely upon broadcasters programming.
Information services, such as those available on the Internet will become increasingly multimedia oriented. The Broadband network will permit high bandwidth surfing from the comfort of your own home.
Other multimedia applications will become possible. Telemedicine, based upon remote access of high resolution medical images, will enable doctors to access medical records such as X-rays or CAT scans from a large city hospital whilst performing surgery in the field. Telepublishing will allow the creation of an individualised electronic newspaper tailored to a persons particular tastes.
Picture this scenario; you are working from home on a project and need to consult with a colleague located in another city. The video telephone allows you to meet virtually face-to-face. You may want to involve more people, so a videoconference can be setup up. You can exchange high resolution images of the design. Comparisons with other designs can be made through image retrieval from a central database. All this is possible without having to leave your own home.
Many of the ideas discussed are already a commercial reality. The technology is already being deployed for such services. The next section will examine the technology that is making Broadband-ISDN a reality.
Understanding ATM Technology
Any technology that is chosen to implement the Broadband-ISDN network must be capable of supporting existing services, as well as future Broadband services. Current networking technologies suffer from drawbacks including:
ATM based networks provide solutions to these problems in existing systems, and also provide the additional capacity and the features to support future high speed applications.
The key components of ATM networks are:
ATM Cell Structure
In ATM, all information to be transferred is packed into fixed sized elements called cells. In this discussion, only the ATM User to Network Interface (UNI) cell is discussed.ATM cells are 53 octets in size of which 48 octets are used to transfer the data payload. The remaining 5 octets comprise a packet header which contains transmission information, data identification and management information. The small size of ATM cells allows rapid, hardware based, analysis and switching.
The ATM UNI cell consists of the following fields:
VPI Field – 8 bit Virtual Path Identifier (VPI) field which is part of the ATM connection identifier and is used for routing. It identifies a group of virtual channels carried along the same route.
VCI Field – 16 bit Virtual Channel Identifier (VCI) field which is also part of the ATM connection identifier. Together, the VCI and VPI fields create point to point connections.
PTI Field – 3 bit Payload Type Identifier (PTI) field which specifies whether the data payload contains management information or user data.
CLP Field – 1 bit Cell Loss Priority (CLP) field which indicates if this cell can be discarded if it encounters network congestion.
HEC Field – 8 bit Header Error Control (HEC) field which permits an ATM network element to determine if the 5 octet header is in error.
Data Field – 48 byte Data Payload field which contains the actual data to be transferred.
STM Data Identification
In conventional Synchronous Transfer Mode (STM), such as Time Division Multiplexing, the position of the data channel determines the identification of the data. In STM systems bit rate allocation is restricted to the predefined channel bit rates.
ATM Data Identification
In ATM systems, the ATM header determines the identity of the data and not the sequential time slot position in a data chain. Consequently, ATM cells may be filled and transmitted in accordance with the user’s instantaneous real needs. This results in a highly flexible data transmission medium.Each ATM cell is a self contained unit and the user’s total information is contained in a collection of one or more cells. Cell allocation is related to users requirements.
Cells are allocated to transport users raw data at a rate determined by the user’s required services (voice, video or data). This key feature gives rise to the term bandwidth available on demand and allows ATM based systems to make optimum use of the available bandwidth, when shared by many users.
Asynchronous Transfer Mode (ATM)
As the VPI/VCI fields are small and hierarchical, the task of forwarding cells is greatly simplified. Because ATM headers are so small it is possible to implement the forwarding decision in a silicon state machine.
Silicon state machines can operate at the wire speed of the output ports (155 Mbit/s or 622 Mbit/s) therefore, the forwarding decision no longer needs to be a bottleneck for network throughput.
ATM switches contain routing tables which are filled with VCI/VPI data gathered from ATM cell headers. Hardware switching takes place based upon these routing tables. In ATM, this routing function sets up a path through the network switches.
Determination of the route to the end-node is only performed once, thus routing intelligence is not expended on every piece of data in a transfer stream. ATM switches and end-nodes establish a route through the network, prior to transmitting data. This operation is known as setting up a connection.
Once established, connections enable data to be forwarded much more efficiently than in traditional networks. This reduces the processing load related to route determination and enables high packet throughput and lower latency networks.
ATM Virtual Connections
ATM networks operate by establishing virtual connections which are paths through the network for transmitting data. Each virtual connection has a Quality of Service (QOS) metric negotiated prior to the transmission of data. This QOS parameter quantifies the desired carrying capacity (in Mbit/s or cells/sec) for each service, the type of data in the data payload and the priority of the data.Virtual connections cause a reservation of network bandwidth for a given transmission task. There are two types of virtual connections. These are Permanent Virtual Connections (PVC) which are established by network configuration procedures and Switched Virtual Connections (SVC) which are dynamically assigned. These SVCs are set up and torn down as needed by traffic on the ATM network.
SVC transmission sessions reserve only the bandwidth needed for a particular transaction. After the completion of the transaction, the allocation is returned for use by other sessions. Switched Virtual Connections are similar in operation to switched calls in current systems.
Each virtual connection may have different cell flow rates and priorities assigned to it and they may be routed through different parts of the network. The establishment of ATM virtual connections provides prior knowledge about data transfer, as well as the means to grant or defer transmission requests based upon local network conditions.
The act of setting up and tearing down data paths provides a high element of network control and enables the network to efficiently process data transfers.
When services need to be transported across an ATM network, they need to be adapted into virtual connections. At the point where a service enters the ATM network, it must be segemented into ATM cells; a process known as cell segmentation. Different types of adaptation are supported depending upon the characteristics of the services.
During the cell segmentation process, the service is segmented into the information field of the ATM cell, and the appropriate header information added. The cells are then multiplexed into an ATM cell stream and enter the ATM network.
Upon reaching their destination, a virtual connection needs to be translated back into the original service; this is known as cell reassembly. The service is extracted from the information portion of the ATM cell and reformatted into the original service.
In ATM networks, when establishing a connection, all nodes in the data path agree to support a traffic contract. This contract specifies characteristics such as maximum sustainable cell rate, peak cell rate and maximum burst duration. The contract also specifies whether the link will be constant bit rate or variable bit rate and whether it will support connection-oriented data (such as Frame Relay) or connectionless data (such as LAN data).At each ATM switch the VPI/VCI field values in the ATM cell are used to determine the output port to push the cell along to the next node. At each node the VPI/VCI fields are updated before the cell is passed along. Each ATM switch maintains an address translation table which is updated each time a new connection is established or an old one is dropped.
An ATM switch may track network connections and know the grade of service that it has agreed to support. This is important as the ATM switch must know how much of the bandwidth is already committed to existing circuits in order to negotiate any contracts for additional connections.
Services are required to condition the traffic they present to the network so that it adheres to the agreed traffic contract. If these services exceed the contract in some way then their data may be discarded in certain cases.
The ATM switches may monitor traffic on each channel to ensure that each connection does not cause problems by using more than its allocated share of the switch’s resources. These features of ATM allow policing of the network and lead to optimum utilisation of bandwidth capacity.
ATM maintains both the real time data transfer characteristics of circuit switching and the flexible speed assignment of packet switching. These attributes of ATM are achieved by dividing information into small cells and transferring these cells at very high speeds. Consequently, ATM is a method which allows integrated circuit and packet switching.In ATM, any type of information can be transferred at either fixed or variable speeds, according to the characteristics of constant or burst data sources. Information from multiple sources can be multiplexed onto a single ATM line, allowing realisation of an efficient network with optimised bandwidth capacity.
Cell identification is contained in small VCI/VPI header fields. The small size of these fields allows hardware based circuitry to switch cell information at very high speeds. ATM based systems operate at low latency speeds and high data transfer rates.ATM can transport voice, video and computer data at a wide range of different speeds. ATM solves latency variation problems by using small, fixed size cells and by interleaving cells from different sources in a network.
ATM addresses the requirements of existing communications systems and has the additional capacity and features to support future, data intensive applications.
ATM Deployment Topologies
ATM promises to become the technology of choice for all networking in both public and private networks. This however will not happen overnight; a gradual migration will happen. In order to understand how this migration will occur, different network categories need to be defined.
Carrier (Public) Network
Within the public carrier there are two classes of network switching; core switching and edge switching.
The core switch represents the backbone of a carriers network. It provides a point for routing all traffic within a public network. Core switches are generally owned and operated by the primary carriers within a country, although they could be owned and operated by large second tier carriers.
Edge switching represents a point of presence in a public network where traffic levels justify concentration but core switches are overkill. Primary carriers deploy edge switches as feeders into core switches. A second tier carrier will use edge switches to offer a competitive service, often linked together via a primary carrier’s core network.Being public networks, standard interfaces to the network are necessary. International standards are being defined for these interfaces.
Corporate (Private) Network
Within the corporate or private network there are a number of categories of networking depending upon the size of the organisation; these are the desktop, the hub, the campus and the enterprise. They may or may not be present in a particular organisation.
The desktop represents the source and destination of all end user services; whether they be data, voice or video. The end user devices usually sit on the desktop and are connected to the corporate network. Initially these will be ATM Adaptor cards in PCs and workstations.
Hubs represent a concentration point of desktop devices within a local area. These will typically occur within a building or office and offer some form of local switching, as well as connection to the rest of the organisation.
Campus switches and PABXs with ATM interfaces represent the next tier in the private networking hierarchy. Switching occurs at one geographic site which might be distributed over a large premises such as a university or large office building. Typically the wiring will be owned by the corporation on their premises. The campus switch will have WAN access to the public network.
An enterprise represents a multi-sited organisation, possibly located in different national or international sites. A private network, implemented between sites, using facilities rented from a carrier (leased lines etc.) is an enterprise network. Switching for the enterprise could be either centralised or distributed depending upon network architecture. Public WAN connections again will be available from an enterprise switch.Being private networks, interfaces within a corporation, do not have to be standard, only the public wide area network (WAN) interfaces need to be standard.
Phased ATM Deployment
ATM is still a new and evolving technology. A lot of standards still need to be defined. This however has not stopped vendors from developing products with proprietary interfaces for the corporate network.ATM will therefore be deployed in the corporate network first. Initial areas of interest will be the campus and enterprise switches as they represent the areas of greatest performance improvements over existing systems. WAN connection will be via leased line or other public services such as ISDN or Frame Relay.
The key to a smooth ATM migration is backwards compatibility with existing networks. The majority of applications are run over legacy LAN’s (Ethernet and token ring). ATM hubs will start to appear with legacy LAN interfaces and ATM backbones.
As ATM demand grows, carriers will start to offer public ATM services. These will initially rollout in the form of trial networks before being available to the wider community.
Finally, ATM to the desktop will start to occur. As ATM equipment becomes price competitive with legacy LAN interfaces, bandwidth intensive applications will become more affordable and user demand will complete the ATM network migration.
With the ubiquity of ATM technology within the network, internetworking becomes a major concern. Interworking between different vendors’ equipment enables customers the freedom to choose the equipment that best meets their needs.Internetworking is achieved through the adoption of open standards. A number of international bodies are continuing to develop the ATM standards with the cooperation of most of the key ATM vendors. The ATM Forum is a user group which has representation from vendors, carriers and users and is the key driving force within the industry.
Work is progressing on an ATM User – Network Interface (ATM UNI) for permanent virtual circuits (PVC’s), with more work still to be done on signaling for Switched Virtual Circuits (SVC’s) and a Network – Network Interface (NNI).
Service mapping, including LAN emulation and voice, are another area being addressed by the standards bodies and will be necessary for internetworking.
Interoperability labs are also springing up that perform interoperability tests between different vendors equipment.
The growth in communications, both in terms of volume and speed requirements, has been phenomenal. Today’s communications networks cannot deliver the necessary bandwidth for emerging services in a cost effective manner.Diverse Broadband services are emerging that require a Broadband-ISDN network capable of carrying them with sufficient quality. A new technology was required to achieve this; that technology is ATM.
ATM is a fast packet switch technology based upon fixed length cells. This permits cell switching in hardware enabling gigabit bandwidth to be achieved. Different services including voice, video and data can be carried by the same ATM network by specifying different quality of services measures.
ATM is applicable across both private and public networks. Carriers will offer public ATM services once the standards have been finalised. In the meantime, vendors have been producing equipment for the corporate market that address private networking needs. This equipment is appearing first at the campus and enterprise level, with compatibility to existing services of paramount importance. Finally, once equipment prices become on par with legacy systems, we will see ATM reach right to the desktop.
This primer has been intended to introduce the reader to Broadband services, the need for a Broadband-ISDN network to support these services and the enabling ATM technology. It has also covered how such technology will be deployed and some of the issues facing the industry as services rollout.