Layer 2 Vpn Architectures [Electronic resources] نسخه متنی

Figure 5-20. OAM Cell Format

Figure 5-21. OAM AIS and Loopback Cell Format

Figure 5-22. Logical Circut Failure AIS and FERF/RDI Alarms

Figure 5-18. The Loopback Indicator field first bit is set to 1 on the outgoing cell and set to 0 to indicate a looped response. The Correlation Tag field matches the outgoing OAM loopback cell with the received response cells. A successive number of loopback replies not being returned could indicate to the endpoint that the logical circuit should be declared unusable.

Managing Traffic

ATM is most well known for the QoS capabilities that allow it to carry a variety of traffic classes. The following are the four general ATM traffic classes:

Constant bit rate (CBR) Used for real-time traffic that consumes a fixed amount of bandwidth. Typical applications include real-time voice and circuit emulation.

Variable bit rate (VBR) Reserved for applications that consume a variable amount of bandwidth. VBR traffic that requires tightly constrained delay and delay variation is classified as Real Time VBR (RT-VBR). VBR traffic that does not have such delay requirements is defined as Non-Real Time VBR (NRT-VBR).

Available bit rate (ABR) Used for non-timecritical applications that support a flow control mechanism to allow it to adjust the bandwidth used based on ATM network characteristics. This traffic class is applicable to any data traffic applications that can take advantage of this variable bandwidth allowed through closed loop feedback mechanisms.

Unspecified Bit Rate (UBR) Intended for non-realtime applications that are delay tolerant. Typical applications include best-effort data transport.

ATM networks employ numerous traffic management mechanisms to maintain the necessary QoS guarantees for each customer PVC or PVP.

ATM Traffic Policing

One of the methods used to meet those traffic agreements is a feature known as ATM policing or usage parameter control (UPC). ATM policing is a mechanism typically performed on ingress into the ATM network to ensure that the traffic received on a logical connection conforms to the defined traffic parameters for that circuit. If the incoming traffic fails to conform, you can discard the data or tag it with a lower priority.

Like Frame Relay policing, ATM policing can be represented as a leaky bucket model, as shown in Figure 5-23.

In a leaky bucket model, each ATM cell has an associated token whose fate determines whether the ATM cell is considered compliant or noncompliant. The bucket represents the number of tokens that can be stored. If the number of tokens exceeds the size of the bucket, the associated cell is considered noncompliant, and appropriate action, such as discarding or tagging the cell, is performed. The leak rate of the bucket represents the rate at which the tokens are drained from the bucket. If the incoming token rate is greater than the leak rate, the bucket will eventually overflow, and the incoming traffic will be considered noncompliant. More complex traffic-policing contracts use a similar model but employ dual leaky buckets.

The ATM Forum Traffic Management 4.0 standard describes several conformance definitions that determine the type of traffic that is regulated and the action that is performed for compliancy/noncompliancy. Table 5-5 describes the traffic conformance definitions that will be explored in more detail in the following sections: CBR.1, VBR.1, VBR.2, VBR.3, UBR.1, and UBR.2. CBR.1, UBR.1, and UBR.2 can be represented as a single leaky bucket with a leak rate that the peak cell rate (PCR) defines. The VBR.1, VBR.2, and VBR.3 definitions are modeled as a dual leaky bucket, with the first and second bucket leak rate equal to the PCR and sustained cell rate (SCR), respectively. The PCR flow and SCR flow columns define the traffic type that is checked for conformance. For example, CLP (0+1) represents all cells, whereas CLP (0) represents only cells with the CLP bit set to zero. The CLP tagging column defines whether the nonconforming action for that bucket is tagged.

ATM Forum TM 4.0 Spec.

PCR Flow

CLP Tagging for PCR

SCR Flow

CLP Tagging for SCR

CBR.1 Traffic Policing

The two values that define the CBR.1 traffic policing model are the cell delay variation tolerance (CDVT) and the PCR. In this model, the PCR (0+1) is the leak rate for all cells, CLP 0 and CLP 1 marked cells. The CDVT (0+1) is the depth of the bucket, which allows for some variation in the token rate. If the token rate is less than or equal to the PCR (0+1), the tokens will be compliant and the associated cells will be allowed into the ATM network. If the token rate is consistently greater than the PCR (0+1) rate, the CDVT bucket depth will eventually be exceeded and those noncompliant tokens, and their associated cells, will be discarded. Figure 5-24 illustrates this process.

VBR.1 Traffic Policing

Unlike CBR.1, which employs a single leaky bucket model, VBR.1 traffic policing can be modeled as a dual leaky bucket, as shown in Figure 5-25. The first leaky bucket acts like the CBR single leaky bucket with a PCR (0+1) leak rate and a CDVT (0+1) depth. Noncompliant tokens in the first bucket are discarded. All CLP 0 and CLP 1 compliant tokens are then checked against the second leaky bucket whose leak rate is SCR (0+1) and depth is a function of maximum burst size (MBS). Tokens that are noncompliant in the second bucket are discarded. Compliant tokens in the second bucket are allowed into the network.

VBR.2 Traffic Policing

VBR.2 traffic policing is modeled as a dual leaky bucket and operates in a similar manner to VBR.1. The difference between the VBR.2 and VBR.1 models is that the second bucket in VBR.2 only checks CLP 0 cells. The compliant CLP 1 tokens from the first bucket are admitted into the network and are not checked for compliance in the second bucket. Figure 5-26 illustrates these differences in the VBR.2 model.

VBR.3 Traffic Policing

VBR.3 policing, illustrated in Figure 5-27, operates in the same manner as VBR.2 except that noncompliant cells in the second bucket are tagged with CLP 1 and admitted into the network instead of being discarded.

UBR.1 Traffic Policing

UBR.1 uses a single leaky bucket model with a leak rate of PCR (0+1) and bucket depth of CDVT. Noncompliant cells are discarded, whereas compliant cells are admitted into the network. Figure 5-28 shows the UBR.1 policing model.

UBR.2 Traffic Policing

As illustrated in Figure 5-29, UBR.2 operates in the same fashion as UBR.1 except that compliant CLP 0 cells are tagged to CLP 1.

ATM Traffic Shaping

ATM traffic shaping is a QoS mechanism that is typically deployed on egress out of an ATM node or end device used to enforce a long-term average rate for a logical circuit. Unlike ATM traffic policing, in which noncompliant traffic is either dropped or marked to a lower priority, ATM traffic shaping queues nonconforming traffic to restrain data bursts and smooth data rates to comply within the defined traffic contract.

Figure 5-30 illustrates the general concept as a leaky bucket model.

Figure 5-30 defines three parameters:

Sustained cell rate (SCR) The average cell rate that traffic should conform to. This is illustrated in Figure 5-30 as the rate at which tokens are replenished.

Peak cell rate (PCR) The maximum cell rate that the traffic cannot exceed. This is represented in the model as the maximum rate at which tokens can leak out of the token bucket.

Maximum burst size (MBS) The number of cells that the device can transmit up to at the PCR rate. The MBS is the depth of the token bucket.

Similar to Frame Relay traffic shaping, cells are transmitted as long as a corresponding token allowing the transmission is available. Traffic is queued for later transmission if a token is not available. If the incoming rate is less than the SCR, tokens are accumulated up to the MBS depth of the bucket. At some later time, if the incoming rate is bursty and exceeds the SCR for a short time interval, the traffic can use the accumulated tokens to send up to the PCR rate. If the incoming rate continues to exceed the SCR rate, the accumulated tokens will eventually be depleted and the cells will only be able to send at SCR, the token replenish rate. Excess traffic will need to be queued and potentially dropped if the incoming rate does not subside.

This generic model applies differently depending on the nature of the traffic class. VBR defines a PCR, SCR, and MBS and follows the general leaky bucket model. On the other hand, CBR's long-term average rate is defined as its PCR and has some form of transmission priority to meet a strict CDVT based on the nature of the traffic it has to support: real-time applications. UBR PVCs typically are not shaped and burst up to the ATM port rate. However, you can optionally define a PCR to limit the maximum transmission rate. ABR is unique compared to the other traffic classes because of its ability to adapt its traffic rate based on indicators of network congestion states such as EFCI or via RM cells. ABR shaping defines a PCR, a minimum cell rate (MCR), the minimum rate that the PVC can send at, and some additional parameters that define its rate adaptation factors.