This section focuses on the detailed design of the multi-site solution, as well as the application scenario, detailed design, and recommended deployment of the multi-site solution.

Application Scenario of the Multi-Site Solution

In a virtualization scenario, a VPC is allocated to a service system. VPCs isolate different users or service systems to prevent them from affecting each other. With the development of services, compute resources required by service systems are increasing. When the capacity of a DC is exceeded, multiple DCs are required to deploy the service system. In this case, the VPC corresponding to the service system needs to be deployed across DCs. For example, VPCs are allocated based on service security levels, intranet, and DMZ. Services in the DMZ are placed in VPC 1, and intranet services are placed in VPC 2. In the multi-active multi-DC DR scenario, the VPCs are distributed to multiple DCs. As a result, a cross-DC large VPC is formed. This is known as the large VPC scenario.

In addition to the abovementioned scenario, different service systems of the same tenant usually need to communicate with each other. For example, traffic from the intranet and DMZ (which need to communicate with each other) is first transmitted to the DMZ and then to the intranet. The DMZ and intranet need to communicate with each other. Because the two VPCs are deployed across DCs, the VPCs across DCs also need to communicate with each other. This is known as the VPC communication scenario.

Deployment of a Large VPC

In the multi-DC scenario, service VPCs need to be deployed across DCs, as shown in Figure 7.9.

The management scope of each SDN controller is limited. Therefore, in a large-scale DC, multiple SDN controllers need to be deployed in multiple DCs. In some cases, multiple controllers are deployed for a single DC. In this instance, network services of the large VPC cannot be delivered by an SDN controller. In addition, an orchestrator is required to collaborate

Networking of a large VPC

FIGURE 7.9 Networking of a large VPC.

Cross-DC deployment of a large VPC

FIGURE 7.10 Cross-DC deployment of a large VPC.

with SDN controllers of multiple DCs to orchestrate VPCs across DCs. In Figure 7.10, one of the major applications of the multi-site solution is cross-DC deployment of a large VPC.

An orchestrator centrally orchestrates networks in two DCs and between DCs. After orchestration is complete, the orchestrator delivers instructions to the corresponding SDN controller to provision VPC and cross-DC interworking instances. The entire network appears as a large VPC.

In the network solution, an independent VXLAN domain is deployed in each DC and is managed by an SDN controller. DCs communicate with each other through three-segment VXLAN or the underlay network.

Two independent SDN controllers and VXLAN domains on the forwarding plane can be deployed to isolate two DCs’ fault domains. This enables customers to deploy DCs in batches or in modular mode.

VPC Communication

Different services need to communicate with each other, as shown in Figure 7.11. In cross-DC scenarios, communication issues also need to be resolved.

Services are deployed by VPC, so service interworking corresponds to VPC communication, as shown in Figure 7.12. In the multi-site scenario, multiple SDN controllers need to collaborate with each other. The orches- trator can orchestrate VPC communication, collaborate with multiple SDN controllers to configure their own network devices, and enable the logical network between VPCs.

< Prev   CONTENTS   Source   Next >