LAN extension refers to the concept of extending Layer 2 networks beyond the traditional Layer 3 boundaries of a single data center. As discussed in the previous section, the ability to live migrate workloads from one physical data center to another is completely dependent on the ability of the Layer 2 network to exist at the destination data center, and on the virtual machine being able to communicate after migration.

While stretching VLANs and address space across two different physical locations facilitates the type of workload mobility being discussed in this paper, it also presents a number of challenges because providing these types of LAN extensions have a big impact on network design.

Unfortunately, it is not possible to simply allow Layer 2 connectivity between data centers using only Layer 3 technologies because this would have significant consequences on traffic patterns between the two data centers, such as STP, unicast floods, broadcasts, and ARP requests. Therefore, technology must be included that employs extended spanning-tree techniques to avoid loops and broadcast storms, and that understands where an active IP address on the subnet exists at any given time.

For further information, refer to your network equipment’s vendor documentation or seek guidance from a network architect who is able to evaluate all of the requirements and constraints of a specific environment. However, when considering the deployment of a LAN extension across different physical sites, it is important to address as key requirements both STP isolation and end-to-end loop prevention to mitigate significant disruptions likely to be caused by remote data center failure or the propagation of unwanted behavior from one data center to the other.