Introduction to Asynchronous Transfer Mode
As the speed and number of local area networks (LANs) continue their relentless growth, increasing demand is place on wide area packet-switching networks to support the tremendous throughput generated by these LANs. In the early days of wide area networking, X.25 was designed to support direct connection of terminals and computers over long distances. At speeds up to 64 Kbps or so, X.25 copes well with these demands. As LANs have come to play an increasing role in the local environment, however, X.25, with its substantial overhead, is being recognized as an inadequate tool for wide area networking. This has led to increasing interest in frame relay, which is designed to support access speeds up to 2 Mbps. But as we look to the not-too-distant future, even the streamlined design of frame relay will falter in the face of a requirement for wide area access speeds in the tens and hundreds of megabits per second. To accommodate these gargantuan requirements, the latest technology has emerged: asynchronous transfer mode (ATM), also known as cell relay.
ATM is similar in concept to 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. ATM is even more streamlined than frame relay in its functionality, and can support data rates several orders of magnitude greater than frame relay.
In addition to their technical similarities, ATM and frame relay have similar histories. Frame relay was developed as part of the work of ISDN, but is now finding wide application in private networks and other nonISDN applications, particularly in bridges and routers. ATM was developed as part of the work on broadband ISDN, but now finds applications in nonISDN environments, where very high data rates are required.
ATM is a packet-oriented transfer mode. It allows multiple logical connections to be multiplexed over a single physical interface. The information flow on each logical connection is organized into fixed-size packets, called cells. As with frame relay, there is no link-by-link error control or flow control.
Figure 1 shows the overall hierarchy of function in an ATM-based network. This hierarchy is seen from the point of view of the internal network functions needed to support ATM as well as the user-network functions. The ATM layer consists of virtual channel and virtual path levels; these are discussed later in this article.
Figure 1 ATM transport hierarchy.
The physical layer can be divided into three functional levels:
Transmission path level. Extends between network elements that assemble and disassemble the payload of a transmission system. For end-to-end communication, the payload is end-user information. For user-to-network communication, the payload may be call-control signaling information (call setup and call termination). Cell delineation and header error-control functions are required at the end points of each transmission path.
Digital section level. Extends between network elements that assemble and disassemble a continuous bit or byte stream. This refers to the exchanges or signal transfer points in a network that are involved in switching data streams.
Regenerator section level. A portion of a digital section. An example of this level is a repeater that's used to simply regenerate the digital signal along a transmission path that's too long to be used without such regeneration; no switching is involved.
Virtual Channels and Virtual Paths
Logical connections in ATM are referred to as virtual channels. A virtual channel is analogous to a virtual circuit in X.25 or a frame-relay logical connection. It's the basic unit of switching in ATM. A virtual channel is set up between two end users through the network; a variable-rate, full-duplex flow of fixed-size cells is exchanged over the connection. Virtual channels are also used for user/network exchange (control signaling) and network/network exchange (network management and routing).
For ATM, a second sublayer of processing has been introduced that deals with the concept of virtual paths (see Figure 2). A virtual path is a bundle of virtual channels that have the same endpoints. Thus, all the cells flowing over all the virtual channels in a single virtual path are switched together.
Figure 2 ATM connection relationships.
Several advantages can be listed for the use of virtual paths:
Simplified network architecture. Network transport functions can be separated into those related to an individual logical connection (virtual channel) and those related to a group of logical connections (virtual path).
Increased network performance and reliability. The network deals with aggregated (and therefore fewer) entities.
Reduced processing and short connection setup time. Much of the work is done when the virtual path is set up. The addition of new virtual channels to an existing virtual path involves minimal processing.
Enhanced network services. The virtual path is used internal to the network but is also visible to the end user. Thus, the user may define closed user groups or closed networks of virtual-channel bundles.
Virtual Path/Virtual Channel Characteristics
The following are characteristics of virtual channel connections:
Quality of service. A user of a virtual channel is provided with a quality of service specified by parameters such as cell-loss ratio (ratio of cells lost to cells transmitted) and cell-delay variation.
Switched and semi-permanent virtual channel connections. Both switched connectionswhich require call-control signalingand dedicated channels can be provided.
Cell-sequence integrity. The sequence of transmitted cells within a virtual channel is preserved.
Traffic parameter negotiation and usage monitoring. Traffic parameters can be negotiated between a user and the network for each virtual channel. The network monitors the input of cells to the virtual channel to ensure that the negotiated parameters are not violated.
The types of traffic parameters that can be negotiated include average rate, peak rate, burstiness, and peak duration. The network may need a number of strategies to deal with congestion and to manage existing and requested virtual channels. At the crudest level, the network may simply deny new requests for virtual channels to prevent congestion. Additionally, cells may be discarded if negotiated parameters are violated or if congestion becomes severe. In an extreme situation, existing connections might be terminated.
Quality of service, switched and semi-permanent virtual paths, cell sequence integrity, and traffic parameter negotiation and usage monitoring are all also characteristics of a virtual path. There are a number of reasons for this duplication:
This provides some flexibility in how the network manages the requirements placed on it.
The network must be concerned with the overall requirements for a virtual path, and within a virtual path may negotiate the establishment of virtual circuits with given characteristics.
Once a virtual path is set up, it's possible for the end users to negotiate the creation of new virtual channels. The virtual path characteristics impose a discipline on the choices that the end users can make.
In addition, a fifth characteristic is listed for virtual paths:
Virtual channel identifier restriction within a virtual path. One or more virtual channel identifiers, or numbers, may not be available to the user of the virtual path, but may be reserved for network use. Examples include virtual channels used for network management.
For virtual channels, there are four methods for providing an establishment/release facility. One method or a combination of these methods can be used in any particular network:
Semi-permanent virtual channels may be used for user-to-user exchange. In this case, no control signaling is required.
If there is no pre-established call-control signaling channel, then one must be set up. For that purpose, a control signaling exchange must take place between the user and the network on some channel. Hence, we need a permanent channel, probably of low data rate, that can be used to set up a virtual channel to be used for call control. Such a channel is called a meta-signaling channel, since the channel is used to set up signaling channels.
The meta-signaling channel can be used to set up a virtual channel between the user and the network for call-control signaling. This user-to-network signaling virtual channel can than be used to set up virtual channels to carry user data.
The meta-signaling channel can also be used to set up a user-to-user signaling virtual channel. Such a channel must be set up within a pre-established virtual path. It can then be used to allow the two end users, without network intervention, to establish and release user-to-user virtual channels to carry user data.
For virtual paths, three methods are defined:
A virtual path can be established on a semi-permanent basis by prior agreement. In this case, no control signaling is required.
Virtual path establishment/release may be customer controlled. In this case, the customer uses a signaling virtual channel to request the virtual path from the network.
Virtual path establishment/release may be network controlled. In this case, the network establishes a virtual path for its own convenience. The path may be network-to-network, user-to-network, or user-to-user.