Layer 2 Vpn Architectures [Electronic resources] نسخه متنی

Understanding ATM

ATM was developed as a high-speed switching solution to handle a variety of traffic types ranging from bursty data services to delay and jitter-sensitive voice. Instead of using variable length frames, ATM utilizes fixed-length cells to transport data. Like Frame Relay, traffic is carried on logical circuits that are uniquely identified by virtual path and circuit identifier fields in the header of each cell.

ATM is probably most well known for its well-developed QoS support because of its strict traffic class definitions. By utilizing a layered protocol architecture, ATM can transport voice, video, and data on the network. Depending on their traffic characteristics, upper layer protocols are processed according to a set of adaptation rules prior to forming each cell.

As subsequent chapters identify, L2TPv3 and AToM interact with three aspects of ATM:

ATM encapsulation to transport either ATM cells or ATM Adaption Layer (AAL) frames on the pseudowire.

Control management/protocol, such as OAM, to properly reflect attachment circuit and pseudowire state

Traffic management features, such as ATM policing and ATM shaping, to emulate ATM's inherent traffic management capabilities

The next section explores these three aspects of ATM as a reference for later chapters.

Encapsulation

To understand ATM encapsulation, it is necessary to describe the lower layers of the ATM protocol stack illustrated in Figure 5-16 and progress through the ATM stack encapsulation from the ATM Adaptation Layer (AAL) and ATM layer with specific focus on AAL5 and ATM cell formats. The lower layers of the ATM protocol stack essentially consist of the following components:

Physical layer The physical layer is composed of two sublayers:

Transmission Convergence (TC) The TC sublayer handles functions such as cell delineation and error detection/correction. The error detection/correction is accomplished by adding a 1-byte CRC to the ATM cell header.

Physical Media-Dependent (PMD) The PMD sublayer is responsible for medium dependent functions such as bit transmission and electro/optical conversion.

ATM layer From a data plane perspective, the ATM layer handles cell header (4 bytes) generation and removal and VPI/VCI translation.

AAL The AAL, defined in ITU-T I.362 and ITU-T I.363, adapts data from various upper layer protocols into the necessary ATM cell payload. The AAL consists of two additional sublayers:

Convergence sublayer (CS) CS processes data from higher layer protocols into variable length frames known as Convergence Sublayer Protocol Data Units (CS-PDUs).

Segmentation and Reassembly (SAR) sublayer SAR is responsible for segmenting the CS-PDUs into 48-byte payloads for an ATM cell.

The next few sections examine the AAL and ATM layer with specific focus on AAL5 and ATM cell formats.

ATM Adaptation Layer

AAL defines multiple AAL formats depending on the traffic type from the upper layer protocols. They include the following:

AAL1 AAL1 is intended to carry connection-oriented, constant bit rate traffic with specific timing requirements. Typical AAL1 traffic is Circuit Emulation Services (ATM Forum standard af-vtoa-0078.0000), such as transparently carrying DS-1 and E-1 circuits across an ATM core.

AAL2 AAL2 supports payloads that have timing requirements similar to that of AAL1 traffic but that have bursty traffic patterns. Compressed voice and video are examples of AAl2 traffic.

AAL3/4 AAL3/4 supports connection-oriented and connectionless variable bit rate traffic. The primary function of AAL3/4 is to carry Switched MultiMegabit Data Service (SMDS) data.

AAL5 Because of AAL3/4's large overhead and complexity, AAL5 was developed as a simpler and more efficient adaptation layer to carry connection-oriented and connectionless traffic. AAL5 is the main format used today for carrying IP routed and bridged data.

Figure 5-17 shows the CS-PDU and SAR-PDU structure for AAL5 and the processing involved down to the ATM layer for cell header generation.

The CS-PDU is formed by appending a CS-PDU trailer to the CS-SDU. The CS-PDU is composed of the following fields:

Padding The Padding field is added to ensure that the resulting CS-PDU size is a multiple of 48 bytes to present to the SAR function.

Common part convergence sublayer user to user (CPCS-UU) The CPCS-UU field allows upper layer protocols to send information transparently to the AAL5 structure. An example application is FRF 8.1, which uses this octet to transport Frame Relay C/R bit. In other cases such as RFC 2684, "Multiprotocol Encapsulation over ATM Adaptation Layer 5," this field is unused.

Common part indicator (CPI) CPI provides alignment of the CPCS-PDU to 64 bits. The value of this field is set to 0x00.

Length This is a 2-byte field indicating the length of the Payload field.

CRC Cyclic redundancy calculated over the entire CS-PDU minus the 4-byte CRC field.

The resulting CS-PDU is presented to the SAR layer, which segments it into 48-byte SARPDUs. The ATM layer generates a 4-byte header for each SAR-PDU. To correctly identify the last cell forming the original AAL5 PDU, the last cell header's third bit in the payload type identifier (PTI), a field in the ATM cell header, is set. The TC sublayer adds the fifth byte to complete the 5-byte header.

ATM Cell Structure

As discussed in the previous section, the ATM layer and the TC sublayer are responsible for the remaining cell header generation and removal. Depending on the interface type (UNI or NNI), the format of the header is slightly different. Figure 5-18 shows the ATM cell format.

The following are the fields in the ATM cell format:

Generic Flow Control (GFC) The GFC field on the UNI header provides flow control on the particular logical PVC. This field is set to 0 and is not fully standardized.

Virtual path identifier/virtual connection identifier (VPI/VCI) Together, the VPI and the VCI uniquely identify a virtual connection. You can use them together as a switching identifier. Alternately, you can use the VPI alone as a switching field and consider it a logical grouping of the VCIs in that scenario. Figure 5-19 illustrates the difference between VP and VC switching.

PTI PTI is composed of 3 bits that characterize the type of cell and measure congestion. The first bit indicates whether the cell is a management cell (1) or contains user data (0). The remaining two bits are interpreted differently in each of those cases, as follows:

User data cell The second bit, known as the explicit forward congestion indication (EFCI) field, indicates congestion. The third bit is set to indicate whether this is the last cell in an AAL5 frame.

Management cell The second bit identifies the cell as an OAM cell (0) or a resource management (RM) cell (1). The third bit distinguishes the OAM cell as an F5 (OAM cell used to convey PVC status) segment (0) or F5 end-to-end flow (1).

Cell loss priority (CLP) The CLP bit prioritizes the cell. In congestion scenarios in which it is necessary to drop traffic, some devices could implement a selective discard mechanism whereby CLP set cells would be dropped before cells without CLP marking.

Header error control (HEC) The TC adds the HEC field, which provides error detection and optionally bit error correction. It is calculated only for the ATM cell header.