New voices gather in the Packet Pushers virtual boardroom for a discussion of Cisco’s layer 2 extension technology, Overlay Transport Virtualization (OTV). Ethan Banks hosts a recording of about two hours worth of content about OTV; this show is the first hour. Joining Ethan are first-time guests Jamie Caesar, Colby Glass and Ken Matlock. Jamie, Colby and Ken have all done real-world OTV deployments. Among the three, both the Cisco Nexus 7000 platform and ASR1000 platforms are represented.
What’s OTV and why do you care? OTV is a layer 2 extension technology aka data center interconnect (DCI). By L2 extension, we mean extending a VLAN from one data center into a different data center, when those data centers are separated by a layer 3 boundary. OTV is a tunneling overlay that encapsulates Ethernet frames so that they can cross the layer 3 area separating the 2 data centers.
While simple in concept and fairly simple to deploy, OTV is rather complex behind the scenes, which is why this was a 2 hour discussion instead of 20 minute one.
Here in part one, we cover the following high level topics.
- What is OTV, and what problems does it solve?
- OTV use-cases beyond vMotion.
- How latency introduced by long-distance DCI impacts applications.
- What hardware can you run OTV on?
- How does OTV compare to other L2 extensions?
- What sort of datagrams does OTV encapsulate?
- How many disparate data centers can OTV stitch together?
- OTV’s fault-isolation mechanisms.
- How is spanning-tree handled in an OTV deployment?
- Is 2 OTV edge devices always an appropriate topology?
- OTV design considerations, hardware and licensing.
- OTV terminology.
- Can you run multiple overlays?
- How do OTV endpoints discover one another when using multicast? Unicast?
Cisco OTV White Papers (excellent resources)