1.3.1 Generic Routing Encapsulation Abstract

In data center applications, there can be a requirement for encapsulation of one protocol over another protocol, for example, in transporting IP over IP for the application of policy. Generic routing encapsulation (GRE) provides what might be considered a lightweight encapsulation mechanism for use in policy and other traffic control routings.

Network Virtualization with GRE decouples virtual subnetworks and addresses from physical network infrastructure, thus providing isolation and concurrence between multiple virtual networks on the same physical network infrastructure.

At the most basic level, a system that uses GRE will have one or more packets that need to be encapsulated and routed. In NVGRE, the encapsulation layers consist of the following:

• Layer 1 encapsulation — a Payload frame that contains an inner Ethernet header and an inner IP header with an IP payload.

• Layer 2 encapsulation — a GRE frame that contains a GRE header ([RFC2890]) that includes a virtual subnet ID. The VSID specifies a hexadecimal value that provides the means of locating virtual machines (VMs) in one or more subnets that are layered upon a physcial network infrastructure.

• Layer 3 encapsulation — a Delivery frame that contains the GRE header, which is encapsulated in the outer Ethernet header and outer IP header.

  The fully encapsulated NVGRE packets can then be forwarded. A simple overview of the fully encapsulated NVGRE packet is represented in the figure that follows.

  

  Figure 1 NVGRE packet encapsulation