Overview

ATM is a high-speed networking technology that handles data in fixed-size units called cells. It enables high-speed communication between edge routers and core routers in an ATM network.

ATM Interfaces

An ATM port can have a major interface and one or more subinterfaces. An ATM subinterface is a mechanism that allows a single physical ATM interface to support multiple logical interfaces. Several logical interfaces can be associated with a single physical interface.

ATM subinterfaces meet the specifications in RFC 2684—Multiprotocol Encapsulation over ATM Adaptation Layer 5 (September 1999), which replaces RFC 1483. All references to ATM subinterfaces in this chapter are still to ATM 1483 subinterfaces.

You can configure a single IP interface on each of the subinterfaces via the interface atm slot/port command. ATM subinterfaces are identified by user-defined numbers. To select a subinterface, you append a subinterface number to the port-level interface atm command. Table 4 shows examples of how the interface atm command is used.

 Table 4:  ATM Interface Commands  

 CLI Command
 Description

 interface atm 3/0

 Select major interface on slot 3, port 0 


 interface atm 2/1.40

 Select subinterface 40 on slot 2, port 1



 

When you create an ATM 1483 subinterface, you must configure a permanent virtual circuit (PVC). Protocols such as ATM require one or more virtual circuits over which data traffic is transmitted to higher layers in the protocol stack.

Figure 1 shows a typical point-to-point ATM interface column.

Figure 1: ATM Interface Column

ATM Physical Connections

ATM interfaces and subinterfaces support two types of connections—point-to-point and multipoint. The router defaults to point-to-point.

• Point-to-point—Indicates a standard connection; for example, connecting two ATM end stations
• Multipoint—Indicates a single-source end system connected to multiple destination end systems. Multipoint indicates a nonbroadcast multiaccess (NBMA) interface. See ATM NBMA.

Depending on the type of connection you choose, you can specify one or more PVCs on each interface. For a standard point-to-point ATM interface, you configure only one PVC. For NBMA ATM connections, you configure multiple circuits.

ATM Virtual Connections

A virtual connection (VC) defines a logical networking path between two endpoints in an ATM network. ATM cells travel from one point to the other over a virtual connection. An ATM cell is a package of information that is always 53 bytes in length, unlike a frame or packet, which has a variable length. An ATM cell has a cell header and a payload. The payload contains the user data.

The cell header includes an 8-bit virtual path identifier (VPI) and a 16-bit virtual channel identifier (VCI).

An ATM network can have two types of VCs, depending on the addressing used to switch the traffic:

• Virtual channel connection (VCC)
• Virtual path connection (VPC)

Virtual Channel Connection

A VCC uses all the addressing bits of the cell header to move traffic from one link to another. The VCC is formed by joining a series of virtual channels, which are logical circuits uniquely identified for each link of the network. On a VCC, switching is done based on the combined VPI and VCI values.

Virtual Path Connection

A VPC uses the higher-order addressing bits of the cell header to move traffic from one link to another. A VPC carries many VCCs within it. A VPC can be set up permanently between two points, and then switched.

VCCs can be assigned within the VPC easily and quickly. The VPC is formed by joining a series of virtual paths, which are the logical groups of circuits uniquely defined for each link of the network. On a VPC, switching is done based on the VPI value only.

ATM SVCs

In addition to PVCs, ATM supports switched virtual circuits (SVCs). PVCs provide a static connection that is usually set up manually. SVCs are established on demand via signaling. Table 5 shows the differences between PVCs and SVCs.

 Table 5:  Differences Between PVCs and SVCs  

 PVCs 
 SVCs 

 Connected on a permanent basis. Users are charged a flat rate.

 Dynamically connect as needed. Users are charged only for time and resources used.


 Manually configured, permanent connections. Each PVC must be configured at both end systems and on all ATM switches in the network.

 Dynamically signalled. SVCs are configured only on the end systems; they do not require configuration on all ATM switches in the network.


 Provisioned when the connection is set up. Bandwidth and services allocated to the PVC are not available to other applications even when not in use.

 Can request bandwidth and ATM service quality information needed for a particular connection. Once the connection is released, network resources are made available to other users or applications.


 Cannot take alternate routes in the event of a failure in the network.

 Can take alternate routes in the event of a failure in the network.



 

Addressing on SVCs

Each ATM endpoint must have a unique address. The ATM Forum defines different types of ATM addresses for use with private ATM networks. These addresses are 20 octets long and are called ATM end system addresses (AESAs). Public networks typically use E.164 native addresses.

For private networks, E-series router SVCs support the three types of AESA addresses shown in Figure 2. For public networks, E-series routers support E.164 native addresses.

Figure 2: AESA Address Formats

Table 6 describes the fields in these addresses.

 Table 6:  Field Descriptions for AESA Addresses  

 Field
 Description

 AFI (authority and format identifier) 

 Identifies the type of address (DCC, ICD, or NSAP encapsulated E.164 address) and the format of the address. The AFI for DCC is 39, the AFI for ICD is 47, and AFI for an NSAP encapsulated E.164 address is 45. 


 DCC (Data Country Code)

 Identifies the country. ISO assigns this code, and each country has a unique DCC.


 DSP (domain-specific part) 

 Contains the actual routing information in three elements: HO-DSP; ESI, which is the MAC address; and the Sel field.


 ESI (end system identifier)

 Identifies the ATM device attached to the switch. Each device attached to the switch must have a unique ESI value (6 bytes).


 E.164 

 Specifies the E.164 address (8 bytes).


 HO-DSP (high-order domain-specific part) 

 Combines the routing domain (RD) and area identifier (Area) portion of the address.


 ICD

 Identifies an international organization and indicates to which code set or organization the particular ICD is assigned.


 IDI (initial domain identifier)

 Identifies the address allocation and administrative authority.


 IDP (initial domain part)

 Identifies the type and format of the IDI.


 Sel (selector)

 Identifies the process within the device that the connection targets (1 byte).



 

SVC Call Setup and Teardown

SVCs use Q.2931 user-to-network interface (UNI) signaling to establish and release switched connections. Before you create SVCs, you configure a PVC signaling channel that the router uses to pass signaling messages during call setup and teardown.

To set up an ATM SVC connection:

1. An end user initiates a call setup request message and forwards the request to the network to establish a connection.

The call setup request message includes information, such as traffic category and service quality information, needed to define and support the connection.

2. The network routes the call setup request through the network to the destination. While routing the connection, the network makes sure that the path has the resources to support the connection information specified in the call setup request.
3. The destination receives the setup request message and either accepts or rejects the call by sending a connect message back to the initiator. The network can also verify the address prefix of the initiator so that the destination can accept or reject the call on that basis as well.

Once established, the connection is available for use until either the source or destination issues a call release message and the end user who receives the message acknowledges the message.

ATM Adaptation Layer

The ATM Adaptation Layer (AAL) defines the conversion of user information into cells by segmenting upper-layer information into cells at the transmitter and reassembling them at the receiver. AAL1 and AAL2 handle intermittent traffic, such as voice and video, and are not relevant to the router. AAL3/4 and AAL5 support data communications by segmenting and reassembling packets.

Your router supports the following AAL5 encapsulation types as specified in RFC 2684—Multiprotocol Encapsulation over ATM Adaptation Layer 5 (September 1999), which replaces RFC 1483:

• aal5snap—LLC/SNAP
• aal5mux ip—VC-based multiplexing
• aal5autoconfig—LLC/SNAP or VC-based multiplexing (see Chapter 13, Configuring Dynamic Interfaces)
• aal5all—Martini encapsulation
  
  

Local ATM Passthrough

E-series routers support local ATM passthrough for ATM layer 2 services over Multiprotocol Label Switching (MPLS). Local ATM passthrough enables the router to emulate packet-based ATM switching. The ATM passthrough feature is useful for customers who run IP in most of their network but still have to carry a small amount of native ATM traffic.

Local ATM passthrough uses ATM Martini encapsulation to emulate ATM switch behavior. You can create pairs of cross-connected ATM VCs within the router. The router then passes AAL5 traffic between two VCs, regardless of the contents of the packets.

You can also use AAL0 encapsulation when you configure a local ATM passthrough connection. AAL0 encapsulation causes the router to receive raw ATM cells on this circuit and to forward the cells without performing AAL5 packet reassembly.

For more information, see JUNOSe Routing Protocols Configuration Guide, Vol. 2, Chapter 5, Configuring Layer 2 Services over MPLS.

VCC Cell Relay Encapsulation

E-series routers support virtual channel connection (VCC) cell relay encapsulation for ATM layer 2 services over MPLS. VCC cell relay encapsulation is useful for voice-over-ATM applications that use AAL2-encapsulated voice transmission.

VCC cell relay encapsulation enables the router to emulate ATM switch behavior by forwarding individual ATM cells over an MPLS pseudowire (also referred to as an MPLS tunnel) created between two ATM VCCs, or as part of a local ATM passthrough connection between two ATM 1483 subinterfaces on the same router. The E-series implementation conforms to the required N-to-1 cell mode encapsulation method described in the Martini draft, Encapsulation Methods for Transport of ATM Over MPLS Networks—draft-ietf-pwe3-atm-encap-07.txt (April 2005 expiration), with the provision that only a single ATM virtual circuit (VC) can be mapped to an MPLS tunnel.

For more information, see JUNOSe Routing Protocols Configuration Guide, Vol. 2, Chapter 5, Configuring Layer 2 Services over MPLS.

Traffic Management

The OC3/STM1 ATM and OC12/STM4 ATM line modules support the following traffic management rates:

• UBR with or without a peak cell rate (PCR)
• VBR-RT
• VBR-NRT
• CBR

The level of support for traffic management is dependent on the specific I/O module. See Supported Modules and Features.

Connection Admission Control

ATM networks use connection admission control (CAC) to determine whether to accept a connection request, based on whether allocating the connection's requested bandwidth would cause the network to violate the traffic contracts of existing connections. CAC is a set of actions that the network takes during connection setup or renegotiation.

The router supports CAC on PVCs on major ATM interfaces. This implementation of CAC determines available bandwidth based on port subscription bandwidth. The router maintains available bandwidth for each major ATM port. Bandwidth for VP tunnels is included in CAC computations.

Table 7 shows the traffic parameter that the router uses for each service category to compute the bandwidth that the connection will require. For example, the peak cell rate is used to calculate how much bandwidth is required for CBR connections.

 Table 7:  Traffic Parameters Used to Compute Bandwidth 

 Service Category
 Traffic Parameter Used to Calculate Required Bandwidth

 CBR

 PCR


 VBR-RT

 SCR


 VBR-NRT

 SCR


 UBR

 UBR bandwidth configured on the ATM major interface


 UBR with PCR

 UBR bandwidth configured on the ATM major interface



 

How CAC Works

With no connections, the available bandwidth is equal to the subscription port bandwidth. As connections are requested, the required bandwidth, which is based on the service category and traffic parameters of the connection, is compared against the available port bandwidth. If sufficient bandwidth is available, the router accepts the connection and updates the available port bandwidth accordingly.

Similarly, when a connection is deleted, the available port bandwidth is updated accordingly.

Configuring CAC

You enable and configure CAC on an ATM major interface using the atm cac command. When you enable CAC on an ATM interface, you can optionally specify a subscription bandwidth and a UBR weight:

• The subscription bandwidth can be greater than the effective port bandwidth to allow oversubscription. The default value of the subscription bandwidth is the effective bandwidth of the ATM port.
• The UBR weight allows you to limit the number of UBR connections by allowing a bandwidth or weight to be assigned to each UBR or VBR with a PCR connection.

ILMI

ATM interfaces support the ATM Forum integrated local management interface (ILMI), versions 3.0, 3.1, and 4.0. An important feature of ILMI is the ability to "poll" or send keepalive messages across the UNI. ATM interfaces always respond to such messages, which are sent by an ATM peer device. Optionally, you can configure ATM major interfaces to generate keepalive messages, a process that enables a continuous ATM-layer connectivity verification; if the ATM peer stops responding to keepalive messages, the router disables the ATM interface.

The ATM interface is not reenabled until the keepalive message's responses are received (or until the keepalive feature is disabled on the ATM port). To enable ILMI and control the generation of keepalive messages, use the atm ilmi-enable and atm ilmi-keepalive commands.

VPI/VCI Address Ranges

The VPI/VCI address ranges allowed on ATM interfaces are module dependent. Certain moduleshave a fixed allocation scheme, whereas others have a configurable allocation scheme. In the configurable allocation scheme, a bit range is shared across the VPI and VCI fields.

For example, if an ATM interface has a bit range of 18, and 4 bits are allocated to the VPI space, then 14 bits are left for the VCI space. The resulting numeric range is 0 to 2ⁿ-1, where n is the number of bits for each space. Completing the example, if 4 bits were allocated for the VPI space and 14 for the VCI space, the configurable range would be 0 to 15 for VPI and 0 to 16,383 for the VCI space. To configure the bit range, use the atm vc-per-vp command.

See Supported Modules and Features for details on how various line module and I/O modules support configurable VPI/VCI address ranges.

VP Tunneling

Virtual path (VP) tunneling allows traffic shaping to be applied to the aggregation of all VCs within a single VP. Thus, VP tunnels can be used to ensure that the total traffic transmitted on a VP does not exceed the specified PCR. VP tunneling uses a round-robin algorithm to guarantee fairness among all of the VCs within the tunnel.

It is possible to change the PCR associated with a tunnel even when VCs have already been configured on the tunnel. The individual VCs within a tunnel must be specified as UBR VCs. In other words, they may not have their own traffic-shaping parameters.

The level of support for VP tunneling is dependent on the specific I/O module. See Supported Modules and Features for details.