This “self-forming” network is based on Extreme Networks gear. The complete solution is called Extreme Automated Campus and consists of several key components:
- switches running the Fabric Engine operating system (formerly VOSS), also known as Universal Hardware. These switches can run either Fabric Engine or EXOS.
- access Points with a feature called Fabric Attach.
- software called ExtremeCloudIQ Site Engine, which provides management, network access control, and traffic analytics.
Core Components of the Solution
The foundation of this solution is the switches running Fabric Engine/VOSS, which can communicate using standard protocols as well as Fabric protocols. Fabric uses only two protocols: SPBM (with MAC-in-MAC encapsulation) and IS-IS as the control plane protocol. The switches form IS-IS adjacencies with each other, but in this context, IS-IS operates at Layer 2. This means that IS-IS programs MAC address tables rather than routing tables, as it traditionally would.
Key Concepts and Terminology
Each switch has its own identifier, known as a BMAC (Backbone MAC Address). In a Fabric network, there are two types of services:
- L2-VSN: Comparable to a traditional VLAN
- L3-VSN: Comparable to a traditional VRF
- I-SID: A service identifier for L2 or L3 VSN
Before diving into how this works and the benefits it offers, let’s clarify a few more terms:
- Customer VLAN (CVLAN): A traditional VLAN.
- Customer MAC Address (CMAC): A traditional MAC address.
- User Network Interface (UNI): An interface to a traditional network.
- Network to Network Interface (NNI): An interface between Fabric nodes.
- Backbone Edge Bridge (BEB): A switch that communicates with both the Fabric and traditional networks.
- Backbone Core Bridge (BCB): A switch that communicates only using Fabric protocols.
How Communication Works: A Simple Scenario
Let’s explore a simple scenario to illustrate how communication works:
The orange section represents the Fabric network with SPBM and IS-IS, while the yellow section represents the traditional network. Host A, located in VLAN 20, wants to communicate with Host B, which is also in VLAN 20:
- BEB A receives a packet from VLAN 20 with the destination MAC address of Host B. VLAN 20 is mapped to I-SID 100 (an L2 VSN).
- BEB A looks into its MAC address database in I-SID 100 and determines that the data should be sent to BEB D. This information is automatically propagated via IS-IS throughout the entire Fabric when IS-IS adjacencies are established.
- BEB A then checks the Fabric tree to determine the path to BEB D—again, information provided by IS-IS. BEB A knows that the next switch in the path is BCB B. The packets are encapsulated, and additional headers (I-SID and the destination MAC address, B-MAC of BEB D) are added.
- BCB B receives the packet and simply switches it to BCB C—the next hop on the path to BEB D, as programmed by IS-IS.
- BCB C does the same, forwarding the packet to BEB D.
- BEB D receives the packet and reverses the process comparing to BEB A — the packets from I-SID 100 are mapped to VLAN 20, the Fabric headers are removed (decapsulation), and the packets are sent to the destination MAC address – Host B.
Benefits of Fabric Connect
The benefits of Fabric Connect are numerous, including:
- support for every topology—ring, spine-and-leaf, star. There are no loops, and all links are active.
- simplified service configuration—L2 VSN and L3 VSN services are configured at the edge of the Fabric and automatically propagated by IS-IS. For example, if you need to configure a new VLAN, you only need to do so at the edge. This can be done manually or automatically.
Fabric can be configured manually, but it can also form automatically when switches running Fabric Engine are connected to each other. I’ll delve into the details in my upcoming blog posts.