A communication protocol, based on small-sized cells, multiplexed logical (virtual) communication channels, and fast packet switching. This protocol is especially suited for constant-bit-rate data. ATM operates over physical layer protocols (like SONET and SDH) that provide switched point-to-point connections that carry constant-bit-rate virtual channels at a wide variety of data rates. See also ATM cell and virtual channel.

Also called ATM network address or ATM hardware address. This address is used within the ATM signaling protocol to identify an endpoint on a switched virtual circuit (SVC). The address is globally unique. Two types of addresses qualify as ATM addresses: ATM NSAP and native E.164. The ATM User-Network Interface Specification, versions 3.0 and 3.1, specify that this address must be assigned to the endpoint by its adjacent switch, using the ILMI address registration procedure. With each permanent virtual circuit (PVC), each ATM endpoint is identified by a local VC address, not by a global ATM network address. See ILMI address registration procedure, ATM NSAP address, native E.164 address, and VC address.

A hardware or firmware module that converts between upper (application) layers and the ATM layer. For transmission, the AAL module creates ATM cells from the upper-layer packets, frames, or bit-stream; this function is commonly called segmentation. On reception, the AAL module converts the ATM cells into a format that is appropriate for the upper-layer application; this function is commonly called reassembly. Currently, there are 5 types of AALs: AAL1, AAL2, AAL3, AAL4, and AAL5.

The basic transmission unit (building block) for ATM. Each cell has 5 bytes of header and 48 bytes of payload (for example, user data or higher-layer overhead), as illustrated in Figure C-1.

The term used in RFC 1577 for ATM address. See ATM address.

A value called in the ATM cell header that identifies one virtual channel (either permanent or switched). The address is composed of a virtual path identifier (VPI) and virtual channel identifier (VCI). See virtual path identifier (VPI) and virtual channel identifier (VCI).

A term used in ATM standards documents for ATM address. See ATM address.

A point of attachment (an interface) between two ATM devices that involves the use of the ATM NNI protocol. In general, an NNI exists at each physical connection within the boundaries of an ATM network (for example, between two ATM switches within a private ATM network). For comparison, see ATM user-network interface (UNI).

A type of ATM address. The ATM NSAP address uses a three-part (that is, AFI, IDI, and DSP) format and abstract syntax that is similar to the OSI network service access point address (OSI NSAP). However, ATM NSAP formats are not OSI NSAPs. All ATM NSAP addresses have a length of 20 octets, and the values for the fields are defined in the ATM User-Network Interface Specification. The ATM NSAP address supports three different address formats. For all formats, the AFI and IDI fields of the address are encoded in binary coded decimal (BCD) format. Each endpoint registers its value for the lower-order DSP portion of the address with its adjacent switch using the ILMI address registration procedure; the endpoint's switch assigns the value for the network prefix portion of the address. The format for the ATM NSAP address is illustrated in Figure C-2; the valid values for the first field of the address are summarized in Table C-1.

Also called, ATM User-Network Interface (UNI) signaling. (This definition does not describe ATM Network-Network Interface (NNI) signaling.) A protocol used between an endpoint and its adjacent switch, whose purpose is to manage and set up a switched virtual circuit (SVC) in real-time. ATM signaling communication occurs on a permanent virtual circuit (PVC) using the VPI/VCI value of 0/5. The protocol specifies a set of communications that allow the following sequence of events to take place:

Signaling is specified by the ATM User-Network Interface Specification and the ITU-T Q.2391 standard; it functions as a layer 3 entity. The protocol is used (“spoken”) over one publicly-defined (that is, well-known) permanent virtual circuit (PVC) in order to set up a system's switched virtual channels. Among other things, the protocol allows endpoints to open and close connections, negotiate traffic contracts, and add and remove parties from multicast connections.

A point of attachment (an interface) between two ATM devices that involves use of the ATM UNI protocol. In general, a UNI exists at each point of entry into an ATM network (for example, between an endpoint and a switch or between a private ATM switch and a public ATM switch). For comparison, see ATM network-network interface (NNI).

A characteristic of a virtual channel whereby cells are placed on the physical medium and managed during their transport in a manner that uses bandwidth that is not being used for constant bit rate (CBR) or variable bit rate (VBR) traffic. ABR is intended for bursty traffic (for example, electronic mail or file transfers).

See transport class.

A notation for representing decimal numerals 0 to 9 using 4-bit binary sequences. The valid binary sequences range from 0000 (for decimal zero) to 1001 (for decimal nine). Each octet (8-bit byte) represents two decimal numerals. For example, the octet 0100 0111 represents 47 decimal. For comparison, 47 decimal in standard binary format, is represented with the following binary sequence: 0010 1111.

The maximum number of cells that a service user is allowed to transmit at the virtual channel's peak cell rate (PCR). Said another way, the longest burst an application can make on a variable bit rate (VBR) channel. This is one component of a variable bit rate VC's traffic contract. See also peak cell rate (PCR) and sustainable cell rate.

Also known as jitter. A performance criteria that measures how consistently ATM cells on a virtual channel arrive at the receiving endstation. For example, if a VC's traffic contract specifies that cells arrive every 110 microseconds, the CDV for that connection increases every time a cell arrives either earlier or later than expected. For constant bit rate traffic, even small CDV values can affect the perceived quality of the data transfer. CDV can be affected by variation in the any of the stations that transmit the cell along the entire connection.

This error is caused when two or more cells (from different VCs) need to be placed into the exact same timing slot within a transmission stream in order to conform to the rates for those VCs. The error occurs on each cell that is not placed in its intended slot.

One-point CDV compares the variability in arrival times in reference to the negotiated peak cell rate (PCR). Two-point CDV measures the variability in arrival times in reference to the time of transmission.

A measurement of the accuracy of ATM data transfer. The result of dividing the number of ATM cells that contain errors by the total number of ATM cells processed. An error can be an invalid ATM header or corrupted payload.

See cell delay variation (CDV).

A 1-bit level of importance indication, carried within the header of the ATM cell. A CLP set to 1 indicates that the cell can be discarded if the network experiences congestion. A CLP set to 0 indicates that the cell should only be discarded as a last resort.

A measurement of the dependability of ATM data transfer. The result of dividing the number of lost ATM cells by the total number of transmitted cells. A lost cell is one that was transmitted but not received within an acceptable period of time.

A measurement of the accuracy of ATM data transfer. The result of dividing the number of misinserted cells by a time interval. A misinserted cell is a cell received on a VC for which there is no corresponding transmitted cell.

A measurement of the speed of ATM data transfer. Cell transfer delay is the elapsed time between when a cell is transmitted and when it is received.

A characteristic of a virtual channel whereby cells are placed on the physical medium and managed during their transport in such a precise manner that each cell arrives at its destination “by the clock”. The time delay between the arrivals of the cells is as close to equal (that is, constant) as physically possible. A CBR channel carries data at its peak cell rate (PCR) all the time. This corresponds to ATM quality of service class A (or 1) and is intended for use by applications such as video-on-demand and telephone conversations. Compare to variable bit rate (VBR).

See native E.164 address and ATM NSAP address.

The number of bits of information that a logical channel carries every second. An embedded transport rate must be lower than or equal to the signal rate of the physical layer.

Also called overlayering. A situation where one network protocol is embedded (encapsulated) within another network protocol's logical structure so that the embedded protocol can be carried transparently (tunneled) through the other network. Figure C-3, illustrates the general header format for an encapsulated packet.

The point at which an ATM service user (original source or ultimate destination) passes the ATM cell payload to the ATM layer or accepts the ATM cell payload from the ATM layer.

A protocol defined in the ATM User-Network Interface Specification, versions 3.0 and 3.1. The protocol is used by an endpoint and its adjacent switch to assign and register a unique ATM address to the endpoint.

An ATM-related protocol that provides status, configuration, address registration, and control information about link and physical layer parameters for an ATM user-network interface. The primary purpose of ILMI is to dynamically discover and register each UNI's ATM address.

Formerly known as CCITT.

A database (for example, a lookup table) that maps IP addresses to ATM addresses. For PVC environments, the table maps IP addresses to VPI/VCI/port tuplets; each endpoint maintains its own table and does its own address resolution. For SVC environments, the table maps IP addresses to ATM NSAP or native E.164 addresses. For RFC 1577-compliant environments using SVCs, the ATMARP server for each IP logical subnet (LIS) maintains this table and responds to requests for address resolution from LIS members. See ATM NSAP address, native E.164 address, and VC address.

See International Telecommunications Union-Telecommunication Sector.

See cell delay variation (CDV).

See signal rate.

A point-to-point physical communication medium. For example, the cable connecting a transmitting station to a switch (or a hub or concentrator) is one link of a longer connection that requires at least one more link in order to reach the receiving station. See also virtual channel link and virtual path link.

A collection of ATM endpoints that all share the same IP network address and mask (that is, the same subnet address) and that can all contact each other directly through the ATM network. The term and its definition are part of the RFC 1577 design for doing IP-over-ATM.

A measurement of the speed of ATM data transfer. An arithmetic average of a specified number of cell transfer delays for one or more connections.

To make one entity serve multiple purposes. In communication environments, this means transmitting multiple communication streams over a single physical medium. For example, 3 conversations can be carried over a single wire by using a different signal frequency for each conversation (an analog multiplexing solution) or by interleaving bytes from the conversations (a digital multiplexing solution).

Also called non-NSAP E.164. A type of ATM address. The format for the native E.164 address is defined in CCITT Recommendation E.164. The addresses are administered and assigned by public networks (for example, telephone companies). This address can be up to 15-bytes in length, as illustrated in Figure C-4.

For ATM NSAP addresses, the network prefix is that portion of the address that is assigned by the ATM switch. This includes all fields in the address except the ESI and SEL fields.

A variable-length address that identifies the location (for example, software module) where an OSI transport layer (Layer 4) accesses the services of an OSI network layer (Layer 3). The NSAP address is defined by OSI documents ISO 8348 and ISO 10589, where values for the different fields are specified and explained. The OSI NSAP format is illustrated in Figure C-5; the valid values for the first field of the address are summarized in Table C-2.

See network service access point address (OSI NSAP).

The maximum transmission rate possible on a virtual channel (VC). For a constant bit rate VC, this is the rate at which the service user transmits data. For a variable bit rate (VBR), this value is the maximum rate that a bursty service user is likely to use. Peak cell rate is one component of each VC's traffic contract.

Functions that exist at various SONET and ATM levels for the purpose of collecting and reporting status that is useful for operating and maintaining a SONET/ATM network.

A measurable characteristic of a communication service. Every protocol uses different performance parameters to define and measure its performance. For the ATM protocol, performance parameters are specified in order to define the quality of service (QoS) classes, as well as to describe and measure performance on a connection. The most important ATM performance parameters are listed below.

Other ATM performance parameters include the following:

See also quality of service (QoS) and the separate entry for each ATM performance parameter.

A long-lived virtual channel that is established through a service order or pre-arranged network management. The traffic contract for this type of channel is negotiated before the service is purchased/installed, and does not change easily, hence the term “permanent.” If any section of a PVC connection fails while the PVC is active, the connection is automatically reopened once repairs have been completed. Contrast this with the switched virtual circuit (SVC).

In general, a specified set of performance parameter objectives that define one level (class) of service. Different parameters and/or objectives for the parameters are grouped into a number of different classes. Customers select the class that fits their needs. As an analogy, for the protocol referred to as overland mail carrier (or U.S. Postal Service), the QoS classes include First, Second, Third, Registered, Certified, and Bulk. Some of the parameters that define these classes are number of days to delivery, level of insurance against content loss or damage, and proof of delivery via a signature at the destination.

In the ATM environment, each virtual channel (VC) has one QoS class associated with it as part of its traffic contract. For a switched virtual circuit (SVC), there is a QoS for each direction (one for forward; one for back/return). A group of VCs that have different QoS classes can be carried together in one virtual path connection (VPC); however, the VPC must meet the requirements for the most demanding QoS among the carried VCs.

The five currently defined ATM QoS classes are listed below. At present, performance parameter objectives for each ATM QoS class vary from provider to provider, and even from contract to contract.

See also performance parameter and traffic contract.

A flow-control protocol associated with the ABR service class.

A measurement of the accuracy of ATM data transfer. The result of dividing the number of ATM cell blocks that contain any of various serious errors by the total number of ATM cell blocks processed. A “block” consists of consecutively transmitted ATM cells that lie between successive OAM cells on a virtual channel connection (VCC).

Also called line rate. The number of bits of information that a physical layer protocol carries every second across the entire length of its physical medium. In synchronous optic network (SONET) environments, the physical medium is a length of fiber-optic cable connecting a SONET transmitter and a receiver. SONET signal rates are usually expressed in megabits per second. See synchronous optic network (SONET).

An ATM adaptation layer (AAL) that supports signaling. See also ATM adaptation layer (AAL) and ATM signaling.

See synchronous optic network (SONET).

Also called STS-1 frame. The basic logical transmission unit used by the SONET protocol. Each SONET frame represents data for one 51.84 megabit per second synchronous transport signal (STS-1). The frame is illustrated in Figure C-6. All SONET signal rates are multiples of the STS-1 rate, and the SONET signal is created by interleaving (multiplexing) bytes from SONET frames. For example, a SONET OC3 (155.52 megabits-per-second) connection carries 3 different STS-1 signals, each using SONET frames to encapsulate its data. The OC3 signal is composed by taking, in turn, the first byte from each signal's first frame, then the second byte from those frames, until all three frames are transmitted, and continuing this process for subsequent frames.

The average transmission rate that a variable bit rate (VBR) service user is likely to use on a virtual channel. This is one component of a variable bit rate VC's traffic contract. See also peak cell rate (PCR) and burst tolerance.

A short-lived, by-demand virtual channel that is established through software negotiation (that is, through ATM signaling). The traffic contract for this type of channel is negotiated at the time the channel is requested. If any section of an SVC connection fails while the SVC is active, the connection is not automatically reopened; the parties involved time out, then reinitiate the connection. Contrast this with permanent virtual circuit (PVC).

An American National Standard (ANSI) protocol for high-speed optical telecommunications. This physical layer protocol is based on optic fiber and synchronous digital multiplexing. SONET supports a wide variety of line rates, starting at 51.84 megabits per second and increasing in multiples of this base rate (for example, 155.52 and 622.08).

The portion of a SONET frame that is provided by the SONET path layer. This area consists of 9 octets of path overhead and 774 octets of payload.

A set of negotiated performance objectives associated with a virtual channel (VC). These objectives are negotiated between a service user and a service provider. The negotiation can be done in any one of three methods: (1) in real time, either through ATM signaling or through a network management system, (2) prior to installation through a verbal or written agreement, or (3) by default (that is, the user accepts whatever performance comes with the service). Real time negotiation via signaling sets up the switched virtual circuit (SVC). The other types of negotiation set up a permanent virtual circuit (PVC).

For private service providers, the traffic contract may consist of any set of parameters.

For public service providers, each traffic contract must include at least the following items:

In addition, the contract may include performance objectives for the following traffic parameters. Not all of these parameters can be specified for all service classes.

See also quality of service (QoS) and performance parameter.

Also known as bearer class. In general, levels of service provided by the entities that transport (that is, bear or carry) the data. These entities are usually referred to as a network. For ATM, there are currently three transport classes, as described below. The exact performance of each class varies from provider to provider because the traffic contract is a negotiated item that can be implemented in more than one manner.

See also quality of service (QoS) and traffic contract.

See ATM user-network interface (UNI).

A characteristic of a virtual channel whereby cells are placed on the physical medium and managed during their transport in a manner that does not require conformance to a traffic contract. Each network provider may define locally a set of performance parameter objectives for UBR, but these do not have to remain constant during the duration of the connection, nor across all links making up the connection. UBR is, by definition, a best-effort transmission, and does not guarantee delivery of the data. This corresponds to ATM quality of service class 0.

A characteristic of a virtual channel whereby cells are placed on the physical medium and managed during their transport in a manner that allows the delay between the arrivals of cells to be unequal (that is, to be variable) even while the bandwidth is guaranteed. The cell delay variation (CDV) remains within the limit specified by the virtual channel's traffic contract. VBR virtual channels are appropriate for bursty transmitters. VBR channels include peak cell rate (PCR), sustainable cell rate, and burst tolerance in their traffic contracts. This corresponds to ATM quality of service class B and C (or 2 and 3) that is intended for applications such as Frame Relay internetworking and compressed video. Compare to constant bit rate (CBR).

The address that uniquely identifies one virtual channel. This address consists of a virtual path identifier (VPI), virtual channel identifier (VCI), and an ATM hardware port identifier. The address is valid only at the ATM station that creates the address. The VC address for each link along an end-to-end connection is different.

Also called logical connection and virtual circuit. A logical communication stream that provides sequential, unidirectional transport of ATM cells in accordance with a traffic contract that is known to both the sender and the receiver. Each virtual channel (VC) is for one conversation, video stream, or other endpoint-to-endpoint communication. Each VC is identified by an address value carried in the header of the ATM cell: this value is the concatenation of the virtual path identifier (VPI) and virtual channel identifier (VCI). VCs come in two varieties--permanent and switched--depending on how their performance parameters (traffic contracts) are negotiated. See also permanent virtual circuit (PVC) and switched virtual circuit (SVC).

A concatenation of virtual channel links that carry the data for a virtual channel. A VCC extends from one endpoint to another; it begins and terminates where the two ATM service users (sender and receiver) access the ATM layer. The collection of physical entities involved in carrying a virtual channel.

A 1- to 16-bit address that in combination with the virtual path identifier (VPI) identifies an active virtual channel. Any specific VCI value is meaningful only along one virtual channel link of the virtual channel connection (VCC); the VCI value in the ATM cell's header is reassigned (and overwritten) at each switch.

A unidirectional physical medium that transports ATM cells between a point where a channel address (that is, VCI) is assigned and a point where the value is translated or removed.

A collection of virtual channels carried together in a single multiplexed stream, all destined for the same terminator (for example, switch). The VP is identified by a virtual path identifier (VPI).

A concatenation of physical virtual path links that extends between two virtual path terminators. The connection carries a virtual path (that is, a bundle of virtual channel). Each VPC has a virtual path identifier associated with it. See also, virtual path link, virtual path terminator, and virtual path identifier (VPI).

A 1- to 8-bit address that identifies an active virtual path (VP). Any specific VPI value is meaningful only along one virtual path link of the virtual path connection (VPC); the VPI value in the ATM cell's header is reassigned (and overwritten) at each virtual path terminator.

A unidirectional physical medium that transports ATM cells between a point where a path address (that is, VPI) is assigned and a point were the value is translated or removed.

A node or system within an ATM network that unbundles the virtual channels contained within a virtual path (VP) and processes each VC independently, which may include rebundling VCs into new virtual paths.

A virtual channel that is reserved for a specific purpose, such as signaling traffic or ILMI communications. The ATM User-Network Interface standard reserves the VCI values (within each VPI) from 0 to 31 for use as well-known channels. The reserved well-known VPI/VCI for ATM signaling traffic is specified as 0/5, and for ILMI communications it is 0/16.