Campus Architecture

For migration purposes, you can add SPBM to an existing network that has SMLT configured. In fact, if there are other protocols already running in the network, such as Open Shortest Path First (OSPF), you can leave them in place too. SPBM uses IS-IS, and operates independently from other protocols. However, as a best practice, eliminate SMLT in the core and eliminate other unnecessary protocols. This reduces the complexity of the network and makes it much simpler to maintain and troubleshoot.

Whether you configure SMLT in the core, the main point to remember is that SPBM separates services from the infrastructure. For example, in a large campus, a user may need access to other sites or data centers. With SPBM you can grant that access by associating the user to a specific I-SID. With this mechanism, the user can work without getting access to confidential information of another department.

The following figure depicts a topology where the BEBs in the edge and data center distribution nodes are configured in SMLT clusters. Prior to implementing SPBM, the core nodes would also have been configured as SMLT clusters. When migrating SPBM onto this network design, it is important to note that you can deploy SPBM over the existing SMLT topology without network interruption. After the SPBM infrastructure is in place, you can create VSN services over SPBM or migrate them from the previous end-to-end SMLT-based design.

Click to expand in new window
SPBM campus without SMLT

After you migrate all services to SPBM, the customer VLANs (C-VLANs) will exist only on the BEB SMLT clusters at the edge of the SPBM network. The C-VLANs will be assigned to an I-SID instance and then associated with either a VLAN in an Layer 2 VSN or terminated into a VRF in an Layer 3 VSN. You can also terminate the C-VLAN into the default router, which uses IP shortcuts to IP route over the SPBM core.

In an SPBM network design, the only nodes where it makes sense to have an SMLT cluster configuration is on the BEB nodes where VSN services terminate. These are the SPBM nodes where C-VLANs exist and these C-VLANs need to be redundantly extended to non-SPBM devices such as Layer 2 edge stackable switches. On the BCB core nodes where no VSNs are terminated and no Layer 2 edge stackables are connected, there is no longer any use for the SMLT clustering functionality. Therefore, in the depicted SPBM design, the SMLT/vIST configuration can be removed from the core nodes because they now act as pure BCBs that simply transport VSN traffic and the only control plane protocol they need to run is IS-IS.

Because SMLT BEB nodes exist in this design (the edge BEBs) and it is desirable to use equal cost paths to load balance VSN traffic across the SPBM core, all SPBM nodes in the network are configured with the same two B-VIDs.

Where the above figure shows the physical topology, the following two figures illustrate a logical rendition of the same topology. In both of the following figures, you can see that the core is almost identical. Because the SPBM core just serves as a transport mechanism that transmits traffic to the destination BEB, all the provisioning is performed at the edge.

In the data center, VLANs are attached to Inter-VSNs that transmit the traffic across the SPBM core between the data center on the left and the data center on the right. A common application of this service is VMotion moving VMs from one data center to another.

The following figure uses IP shortcuts that route VLANs. There is no I-SID configuration and no Layer 3 virtualization between the edge distribution and the core. This is normal IP forwarding to the BEB.

Click to expand in new window
IP shortcut scenario to move traffic between data centers

The following figure uses Layer 3 VSNs to route VRFs between the edge distribution and the core. The VRFs are attached to I-SIDs and use Layer 3 virtualization.

Click to expand in new window
VRF scenario to move traffic between data centers