Maximum potential economic benefit of Diff Serv is when traffic requiring the highest SLA represents a minor proportion of the overall capacity

The figure lists two of the main benefits of using DiffServ over a best-effort with overprovisioning backbone.

Instead of overprovisioning based on the aggregate bandwidth of all traffic, with DiffServ, each traffic class can be designed with different provisioning ratios. If the premium traffic class is only 20 percent of the total capacity, overprovisioning for 100 percent of the traffic load is expensive and not necessary to guarantee the QoS requirements for the premium traffic class.

The following example illustrates an overprovisioning example with and without DiffServ:

If the total aggregate bandwidth is 10 Gbps, where the premium class = 2 Gbps, the business class = 3 Gbps, and the default class = 5 Gbps:

■ With best effort and an overprovisioning ratio of 2:1, the provisioned bandwidth = 10 Gbps

■ With DiffServ, and the premium class having an overprovisioning ratio of 2:1, the business class having a lower overprovisioning ratio of 1.5:1, and the default class not having any overprovisioning, the provisioned bandwidth = (2 Gbps * 2) + (3 Gbps * 1.5) + 5 Gbps = 13.5 Gbps.

By isolating the traffic into different traffic classes, then treating the different traffic classes with different PHBs, DiffServ can reduce the bandwidth requirement on the network while achieving the same SLA when compared to the non-DiffServ case.

PE-to-P QoS PE Outbound

The figure illustrates the typical QoS configurations required at the service provider core PE and P routers.

The complex QoS policies of classification and marking, policing, shaping, LFI, and cRTP are only required at the edge. In the core, only LLQ (or MDRR for GSR) and WRED are needed.

The figure shows the QoS configurations on the ingress PE router outbound interface to the P router.

In this case, a traffic policy called "OUT-POLICY" is configured to provide the LLQ or CBWFQ, and WRED. Each traffic class bandwidth guarantee is configured using a percentage rather than a fixed bandwidth in kbps.

No inbound policy is required on the P router.

P-to-P QoS P Outbound

The figure shows the QoS configurations on the P router outbound interface to another P router.

In this case, a traffic policy called "OUT-POLICY" is configured to provide the LLQ or CBWFQ, and WRED. Each traffic class bandwidth guarantee is configured using a percentage rather than a fixed bandwidth in kbps.

No inbound policy is required on the receiving P router.

P-to-PE QoS P Outbound

The figure shows the QoS configurations on the P router outbound interface to the egress PE router.

In this case, a traffic policy called "OUT-POLICY" is configured to provide the LLQ or CBWFQ, and WRED. Each traffic class bandwidth guarantee is configured using a percentage rather than a fixed bandwidth in kbps.

No inbound policy is required on the egress PE router.

Was this article helpful?

0 0

Post a comment