0

While learning about Fibre Channel and SAN, I found the following information on IBM Knowledge Center:

A switched fabric is a fibre channel topology in which individual node ports are interconnected and managed by switches.

However, AFAIK, the concept of switched fabric consists of interconnecting various nodes via one or more network switches.

From the physical link POV, I think this is possible to do in Ethernet, since it is possible to connect several switches with nodes using Ethernet interfaces, and forwarding is also possible. Although I don't understand, why Ethernet doesn't support the switched fabric topology.

Why this technology is specific to FibreChannel, why not Ethernet as well?

5

"Switched fabric" (SW) is mainly used for Fibre Channel networks as its contrasts the other FC topologies point-to-point (P2P) and arbitrated loop (AL). The point in SW being that there are multiple, active paths between two nodes.

Classic Ethernet requires a tree topology - however, the concept of a fabric isn't uncommon with Ethernet, especially when used with IP.

In an iSCSI storage network (SAN), most often a fabric is built on the IP layer (L3) - because it's easier to build and easier to control. Building a meshed network with active links directly on Ethernet's layer 2 requires either a tricky MSTP setup, multi-chassis link aggregation (MLAG), or Shortest Path Bridging (SPB aka 802.1aq).

In a hierarchical network design, the core and the distribution layer often use fabrics as well. They are shifting towards routed links (L3) instead of switched links (L2) to increase scalability and efficiency, and speed up convergence in case of failure.

So basically, from a certain (high) vantage point there's not even that much difference between a Fibre Channel switched fabric and a meshed, routed Ethernet/IP LAN.

  • Thanks for your answer. So, the only difference between switched fabric and ethernet meshs is the performance, because of the multipathing? – Kais Dec 17 '18 at 22:27
  • For the most part, yes. A fabric is a multi-point construct where multiple paths are expected. A-B-C-A is a problem for ethernet -- something (STP) has to break one of the links -- whereas for FC it's expected and traffic can (and does) flow in both directions. – Ricky Beam Dec 18 '18 at 1:20
  • @Kais With simple Ethernet (switches), there can only be a tree - no redundant paths. With STP and LAG you can add aggregated and failover links, but there's no fully active L2 mesh. Using MLAG or SPB or with IP you can build a fully active mesh. – Zac67 Dec 18 '18 at 7:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.