Asynchronous transfer mode

Asynchronous Transfer Mode ( ATM ) is a communication protocol that is suitable for the transmission of data , voice and video. Layer 2 packets are called cells or slots , have a fixed length (53 bytes, 48 ​​of which are data, 5 bytes of cell header) and are transmitted using asynchronous time division multiplexing .

Overview and history

ATM was developed as a network standard that uses the synchronous ( Plesiochronous Digital Hierarchy (PDH), Synchronous Digital Hierarchy (SDH)) transport techniques and adds other useful features to them. Not only circuit-switched data transmission is supported by ATM, but also packet-based data such as IP , Frame Relay etc. In contrast to the simple and robust Ethernet technology, which can lead to unpredictable results in load situations, ATM offers guarantees with regard to effective bit rate , delay and jitter , what is commonly (among other properties) referred to as Quality of Service (QoS).

The problem of merging diverse data streams of different types, i.e. being able to work with both synchronous and packet-based networks, for example, was solved by converting both bit stream types (synchronous or packet-based) to a new bit stream with ATM cells at the intersections. The cells are typically sent in the payload of PDH- or SDH -formatted data streams. Asynchronous in ATM means that the sender and receiver can work with clock rates that differ from one another within wide limits : The receiver uses the Header Error Check (HEC) to check for each cell whether there is sufficient synchronization and, if necessary, carries out a new synchronization.

In the original concept, ATM was the key technology for the Broadband Integrated Services Digital Network ( broadband ISDN ), which should form the backbone network for the existing analog telephone network ( Plain Old Telephone System , POTS ) . The entire ATM standard therefore consists of definitions for layers 1 to 3 (bit transmission layer, security layer and network layer) of the OSI model . Telecommunications companies, but also the American Department of Defense (DoD), were in charge of developing the ATM standards . As a result, many of the existing telecommunications procedures and conventions have been incorporated into ATM.

Today ATM technology is used to support applications from global Internet and telephony backbones to DSL technology and private LANs . The specifications are developed by the ATM Forum . The specifications are then submitted to ITU-T (formerly CCITT) for standardization.

ATM standards

ATM layer model (levels):

higher layers for user data, control bits
ATM adaptation layer
ATM layer
Physical layer

Management functions ( OAM ) are defined to a much greater extent for ATM than for IP . They include configuration management, fault management, and performance measurement.

Tasks of the shifts:

Physical layer

Physical connection to other systems, preferred medium SDH

ATM layer

Transport and switching of ATM cells

Adaptation Layer or ATM Adaptation Layer (AAL)

The task of the AAL is to adapt data from higher layers to the format of the user data field of the ATM cell and to transmit control information to the opposite end. A distinction is made between five types of service, of which, however, so far only the simplest has a greater significance. IP uses the ATM Adaptation Layer 5 ( AAL5 ). The adaptation of the AAL5 mainly takes on the fragmentation and reassembly for the IP packets that do not fit into the short user data field.

ATM concepts

Why cells?

The reason for using small data “cells” was to reduce jitter when multiplexing data streams.

When ATM was developed, STM-1 lines with 155 Mbit / s (135 Mbit / s payload) were a fast optical network connection, whereby many PDH lines of the networks of the time were significantly slower: 1.544 Mbit / s to 45 Mbit / s in in the USA and 2 to 34 Mbit / s in Europe.

A standard IP data packet of maximum length (1546 bytes / 12368 bits, although the IP specification allows 64 KiB) needs between approx. 90 µs (135 Mbit / s) and 8 ms (1.544 Mbit / s) for transmission with these data rates and blocks the data channel during this time.

If a voice signal divided into packets has to share the line (data channel) with large-volume data traffic, these voice packets - no matter how small they are made - always meet full-size data packets and have to wait a correspondingly long time before they can be transmitted. These delays were too long for voice traffic, so that even after filtering out the jitter, echo cancellation would have been required even in local networks . That was simply too expensive at the time.

The solution to this problem was to split all packets into 48-byte sub-packets, to add a routing header of 5 bytes and then to multiplex these 53-byte cells instead of the original packets. The original packets can later be identified and reassembled using the header. This procedure reduced the queuing time to almost a thirtieth, which saved the need for echo cancellation.

The rules for dividing and reassembling packets and streams in cells are known as ATM Adaptation Layers : The two most important are AAL 1 for streams ( e.g. voice) and AAL 5 for almost all types of packets. Which AAL is used is not encoded in the cell. Instead, it is configured between two endpoints or agreed on the basis of a virtual connection.

Today, a full-length Ethernet packet only needs 1.2 µs on an optical connection with a 10 Gbit / s data transmission rate, which actually makes it no longer necessary to use small packets in order to keep latency times short. Some conclude that ATM has become superfluous in backbone connections.

ATM is still useful for slow connections (up to 2 Mbit / s). For this reason, many ADSL systems use ATM between the physical layer and a layer 2 protocol such as PPP or Ethernet .

Why virtual connections ?

ATM is based on connections that can be set up permanently and can only be switched for a certain period of time using ISDN- like signaling . Virtual paths (VPs) and virtual channels (VCs) have been defined for this purpose . Each ATM cell contains a virtual path identifier (VPI, 8 or 12 bit) and a virtual channel identifier (VCI, 16 bit) in the header . While these cells pass through the ATM network, the switching is achieved by changing the VPI / VCI values. Although the VPI / VCI values ​​do not necessarily stay the same from one end of the connection to the other, this corresponds to the concept of a connection, since all packets with the same VPI / VCI values ​​take the same path, in contrast to IP , in which one packet could reach its destination via a different route than previous and following packets.

Virtual connections also have the advantage that they can be used as a multiplexing layer for different services (voice, frame relay , IP , SNA etc.) which can then share a common ATM connection without interfering with each other.

Traffic management with cells and virtual connections

Another key concept of ATM is called “Traffic Contract”: When an ATM connection is set up, every switch along the way is informed of the traffic class of the connection.

Traffic contracts are part of the mechanism through which Quality of Service (QoS) is implemented. There are four basic types (with several variants), each of which describes a set of connection parameters:

Unspecified Bit Rate (UBR) (German "undefined bit rate")

is the default type for "normal" traffic. This is where you get the bandwidth that is left after the QoS traffic is processed. So this is a best effort connection.

Available Bit Rate (ABR)

the transmission rate is regulated based on the currently free bandwidth. The regulation takes place either via the EFCI flag in the cell header or via special resource management (RM) cells.

Variable Bit Rate (VBR) (German "variable bit rate")

Here you “order” an average cell rate, which you can exceed by a certain amount for a certain period of time (there are real-time (RT-VBR) and non-real-time variants (NRT-VBR)).

Constant Bit Rate (CBR) (German constant bit rate)

here a peak data rate ( Peak Cell Rate , PCR ) is requested, which is then guaranteed. On the other hand, this also means that u. U. bandwidth remains unused.

Compliance with traffic contracts is usually enforced through “shaping”, a combination of queuing and classification of packets, and “policing” (application of “guidelines”).

Traffic shaping

Usually the shaping takes place at the entry point of an ATM network, and there an attempt is made to control the cell flow in such a way that the traffic contract can be adhered to. The simplest form is Peak Cell Rate ( PCR ) Shaping, which limits the maximum cell rate to a specified value. Shaping within the ATM network requires buffer capacities ( Buffer Manager ), as the cells are occasionally forwarded with a delay and therefore cells accumulate.

Traffic policing

In order to increase data throughput, it is possible to assign rules to virtual connections that contradict your traffic contracts . If a connection exceeds its traffic contract , the network can either discard the cells itself or set the Cell Loss Priority (CLP) bit to mark the packets as discardable for further switches on the way. This policing therefore works cell by cell, which can lead to problems if packet-based communication is implemented on ATM and the packets are packed in ATM cells. Of course, if one of these ATM cells is discarded, the entire previously segmented packet becomes invalid. For this reason, schemes such as Partial Packet Discard (PPD) and Early Packet Discard (EPD) were invented , which discard a whole series of cells until the next data frame begins (see Discard protocol ). This reduces the number of redundant cells in the network and saves bandwidth for full data frames. EPD and PPD only work with AAL 5 because they have to evaluate the frame end bit in order to determine the end of a packet.

Structure of an ATM cell

An ATM cell consists of a header of 5 bytes and 48 bytes of user data ("payload"). The useful data size of 48 bytes resulted as a compromise between the needs of voice telephony and those of packet-based networks. One simply took the average of the packet lengths of the American (64 bytes) and the European proposal (32 bytes).

ATM defines two different cell formats: NNI (Network Network Interface) and UNI (User Network Interface). Private ATM connections use the UNI format, public ATM networks use the NNI format.

Diagram of a UNI-ATM cell 7th 4th 3 0 GFC CPI CPI VCI VCI VCI PT CLP HEC Payload 48Byte

Diagram of an NNI ATM cell 7th 4th 3 0 CPI CPI VCI VCI VCI PT CLP HEC Payload 48Byte

( GFC : Generic Flow Control, VPI : Virtual Path Identifier, VCI : Virtual Channel Identifier, PT : Payload Type, CLP : Cell Loss Priority, HEC : Header Error Control)

In a cell with UNI header format, the GFC field is reserved for a local flow control between network and user (which is still undefined today). Because of this planned use, the transmission of GFC bits is not guaranteed by public ATM networks. Until the local flow control is standardized, all four bits must be set to zero by default. In private networks, they can be used as required, unless manufacturer-specific restrictions prohibit this.

The NNI format of an ATM cell is identical to the UNI format except for the missing GFC field. Instead, these bits are used to enlarge the VPI field from 8 to 12 bits. Therefore 2 ¹² VPs with 2 ¹⁶ VCs each can be addressed or the corresponding number of connections can be switched via a single port . In the UNI format there are only 256 VPs with 2 ¹⁶ VCs each . In practice, some VP / VC numbers are usually reserved for special purposes and therefore cannot be used for useful connections.

The PT field is used to differentiate between different types of cells for user data or maintenance and management purposes . So z. E.g. cells for the exchange of signaling information, control data for the monitoring of network elements as well as cells for resource management and traffic control .

The Cell Loss Priority Bit (CLP) indicates whether the cell has a high (CLP = 0) or low priority (CLP = 1). This is only important if a network node is overloaded and some cells have to be discarded. Cells with the low priority are discarded first. The CLP bit can be set or changed by terminal equipment or network nodes.

The HEC field (Header Error Correction, checksum of the header) makes it possible to check whether the header of the ATM cell was transmitted without errors; an error check of the user data must be carried out in higher layers. It is also used for cell synchronization: if the receiving side has not correctly identified the beginning of the cell, it also takes the wrong bytes as the HEC field and then comes to negative test results until it has synchronized itself again to the correct beginning of the cell.

See also: DSS2 , DSL , IP , MPLS , DQDB

Numbering in ATM networks

Originally it was planned that the same number range would be used for the B-ISDN based on ATM technology as for the ISDN , i.e. the one standardized according to ITU-T recommendation E.164 . But after the IT world had recognized ATM technology as usable, a fierce battle against this numbering scheme took place. It ended with the creation of an alternative number space, which is known today as the “ATM End-System Address” (AESA). This prevented the national telecommunications companies, which at that time were still often monopolies, from dominating number allocation. Today, both address types are common, but they are fundamentally different:

1. Address type A consists of an international address in accordance with E.164 with a subaddress. The subaddress contains the necessary information with which the terminal is identified. The subaddress can come from a private namespace and be based on an AESA.
2. Address type B is the AESA, which is based on the format of the OSI NSAP address (but is not itself).

The former debate about ATM addressing is similar to today's debate about addressing in IP telephony , which, among other solutions, has led to what is known as Telephone Number Mapping .

Application and operator

Almost all operators of communication networks have set up ATM networks in the backbone area, but do not use ATM signaling, but fixed connections. In the broadband access network , ATM is used almost exclusively as a multiplex layer ( DSLAM , RAS ). ATM was only able to establish itself as a technology for local networks in the high-performance area. Its high complexity and the associated costs prevented its large-scale use as an integrated network solution in the office area. Some basic principles of the ATM standards, such as the ability to prioritize certain types of data traffic, were later adopted in MPLS , a general protocol for efficient switching below Layer 3.

Use by radio and broadcasting stations

The use of ATM technology is used by broadcasters in Germany: Larger production companies and broadcasters use fiber optic networks to send their audio and video material to the various broadcasters in real time - the ARD internal production network ( HYBNET ) is based on ATM technology . With the help of satellite technology ( uplink ), larger distances (intercontinental) can be bridged via the ATM network. The transfer points consist of an encoder and a decoder , the so-called muxer ( multiplex method ). The technology also provides the basis for switching live from studio to studio.

Current situation

For example, although the classic telecommunications network operators invested enormous sums in ATM infrastructure, there have been increasing indications since the late 1990s that more and more applications are using other, often Ethernet- based technology instead of ATM. Reasons for this could be the much lower prices of devices of the IEEE 802.3 family and the more easily accessible know-how.

The following parts of this section since 2012 seem out of date to be : plans by 2012 to ATM has now done?
Please help to research and insert the missing information .

Deutsche Telekom AG plans to stop upgrading its Internet DSL connections and telephone exchanges with ATM technology by 2012, so that ATM will no longer play a major role in the backbone area in the future. ATM technology is being replaced by Ethernet- based technology and IP-based VPNs. In February 2013 it became known that Telekom intends to gradually switch off “little-used parts of conventional network technology” (meaning ATM DSLAMs ) in a test phase up to 2016 and migrate customers to IP solutions (“All-IP”). Several thousand customers who have already been written to are affected. The switching centers are to be retained, but the network operator wants to "gain important experience for the migration to IP technology, which should then be incorporated into further planning."

Although ATM as a technology will probably no longer play a role, it should also be mentioned that some of the research findings gained with ATM will continue to be used in other network technologies, such as MPLS. But QoS in the Internet or in future networks or TCP congestion control has also benefited from ATM - at least in the area of ​​research (see TCP / congestion control as a research field ).

See also

• ILMI
• ATM25

Web links

Commons : Asynchronous Transfer Mode  - collection of images, videos and audio files

• ATM cell formats

Individual evidence

1. ↑ ATM versus Gigabit Ethernet
2. ↑ ATM versus Ethernet ( Memento from August 24, 2011 on WebCite )
3. ↑ Cisco recommends Customers migrate from ATM to Ethernet ( Memento of March 12, 2007 in the Internet Archive )
4. ↑ Gigabit Ethernet is on the rise
5. ↑ Lucent ATM to Ethernet Migration ( Memento from March 4, 2007 in the Internet Archive )
6. ↑ Telekom: The end of the analog landline connection will come in 2016. Telariff, February 21, 2013
   ( Note: The article title should not be misunderstood in such a way that analog landline connections will no longer be available after 2016; only connections that are over ATM DSLAMs that are still in operation are connected to the fixed network.)