Tweet This: Tweet
Share on LinkedIn:
By Neal O’Horo, Sr. Network Engineer
It’s that time of the year again, another great Cisco Live. I decided to continue my journey from the last year and focus on the spine-and-leaf and IP Clos, here used interchangeably, Virtual Extensible Local Area Network (VxLAN) fabric, providing an introduction and some purposes, and, hopefully, making it easier to understand, using some concepts many network engineers know.
Imagine a modular chassis with some line cards. Instead of the line cards, you have the leaves, of which you may know as the Top of Rack (ToR) or End of Row (EoR) switches. Instead of the Serial Line Interfaces (SLI) on a mid or back plane, you have the Ethernet uplinks. Instead of the mid or back plane, you have the spines, of which you may know has an aggregation/distribution switch, with the Ethernet uplinks on the leaves connected to each of the spines with either a single link or a Link Aggregation Group (LAG). In essence, the IP Clos VxLAN fabric — a.k.a., the spine-and leaf — is akin to a modular switch or set of High Availability (HA) Ethernet switches with some line cards, except it’s distributed amongst many chassis, and the control and management may exist locally or centrally in a cloud or external system.
Some of the other IP Clos fabric concepts may be familiar to you, as well. The VLAN becomes a VxLAN with 24 bits, 0s and 1s, instead of 10 bits, giving you the ability to have more LANs. You continue to use some familiar routing protocols concepts, such as the Multiprotocol Border Gateway Protocol (MBGP), Open Shortest Path First (OSPF), Intermediate System-to-Intermediate System (ISIS), and Equal Cost Multi-Path (ECMP) which can negate the Spanning Tree Protocol (STP) to handle the loop prevention. The LAGs continue to use the Link Aggregation Control Protocol (LACP). The legacy Switched Virtual Interfaces (SVI) and Integrated Routing and Bridging (IRB) interfaces can still be used and are similar to the VxLAN Network Virtual Interfaces (NVE), allowing for the bridging between the different VxLAN segments, known as a Layer 3 gateway. Layer 2 gateways, also exist, allowing for the bridging between a VLAN and a VxLAN. The leaf switches are also known as the VxLAN Tunnel Endpoints (VTEP). The tunnels themselves are known as the UDP tunnels, similar to the Generic Routing Encapsulation (GRE) tunnels. Accordingly, the IP Clos fabric incorporates some other technologies, such as MPLS BGP route-target communities and route-distinguishers and VPN Routing and Forwarding (VRF) instances, each terms and concepts familiar to the network engineers, helping you to understand the technology and the purposes for the IP Clos VxLAN fabric.
Some of the purposes for the IP Clos fabric may be clear. It scales well: If you need some throughput, you add a spine and an uplink. If you need some more ports, you add a leaf. If you want some load balancing or loop prevention, it’s one fabric; these concepts are part of the fabric. If you need a Datacenter Interconnect (DCI), this, too, can be part of the fabric. If a VM moves, the fabric knows where the VM MAC address resides. If you want to manage and control the fabric, this can be abstracted. With the VRFs, if want some multiple fabrics, you can have them. If a chassis fails, the fabric knows a neighbor is down, just like in the normal IP routing. The Anycast Gateway, one IP address for each LAN across the fabric, replaces the legacy Next-Hop Routing Protocols (NHRP); no more does the traffic need to trombone across a DCI to switch between the different LANs or reach the Internet. Importantly, the IP Clos VxLAN fabric can be agnostic; any network OEM should be able to support it. The fabric can function as an underlay and an overlay, existing physically or virtually on the premise or in a cloud. This list continues.
I hope this provides enough information to help understand the basic spine-and-leaf, IP Clos, VxLAN fabric. It uses some concepts of which many network engineers may already understand and use today. A spine-and-leaf is like a modular chassis, except the components are individual chassis and links that form one large virtual chassis or set of virtual chassis that can exist at one site or many. For the next time, I’ll try to expand upon the VxLAN and its routing.
Looking to learn more about modernizing and automating IT? We created the Kovarus Proven Solutions Center (KPSC) to let you see what’s possible and learn how we can help you succeed. To learn more about the KPSC go to the KPSC page.
Also, follow Kovarus on LinkedIn for technology updates from our experts along with updates on Kovarus news and events.