SD-WAN Application Flows and Policies

Forwarding of IP Packets across different Underlay Connectivity Services with different attributes based on Policies applied to Application Flows is one of the defining characteristics of SD-WAN Services.

As part of the SD-WAN Service, the Subscriber and the Service Provider agree on the Application Flows that are identified at the SD-WAN Edges. For each of the agreed-on Application Flows, a Policy (list of Policy Criteria) is assigned, which defines how IP Packets in the Application Flow are handled.

An Application Flow can be described by a broad set of characteristics of the packet stream that can be identified at the UNI[1], including standard layer 2 and layer 3 fields such as addresses, ports, and protocols. In addition, many SD-WAN implementations can perform deep packet inspection (DPI) up through layer 7. Therefore:

• An Application Flow can include IP Packets for several individual computer applications, such as “all packets that use the RTP protocol as defined in RFC 3550 [13]” or, conversely,
• IP Packets for a single application could be split among multiple Application Flows, such as a single video conferencing call resulting in a “Video” Application Flow and an “Audio” Application Flow, or,
• An Application Flow can include all IP Packets from an IP address range such as 10.10.10.x/24, which could, for example, represent all Point of Sale terminals at a location.

Forwarding of an Application Flow is based both on the Policy assigned to the flow and IP forwarding requirements, which together determine the best TVC for forwarding each IP Packet in the Application Flow. Appendix A includes an example of how the implementation of this process could be structured.

[1] The techniques and technologies used to identify the flows are outside the scope of this specification.