Fibre Channel over
Ethernet (FCoE) enables I/O consolidation. It permits both LAN and SAN traffic
to coexist on the same switch and the same wire. This feature enables you to
consolidate multiple separate networks into a single converged infrastructure.
Key values of I/O
consolidation using traditional FCoE are as follows:
separate network infrastructures for SAN and LAN traffic.
hardware requirements, such as cabling and server interface cards (NICs and
HBAs), and lowering capital expense.
Reduction in power
and cooling requirements for fewer physical assets.
deployment agility for multiprotocol networks, which preserves long-term
investments while preparing for future uncertainty in protocol needs.
By using FabricPath
Ethernet technology, you can take FCoE consolidation even further:
Create a logical,
rather than physical, SAN A/B separation.
balance multiprotocol traffic within the data center.
establish relationships between switches, reducing the possibility for human
error during configurations.
availability percentages as the scale increases.
architecture provides an inherent multipath capability with redundancy to
handle node failures. Fabric level redundancy is provided through a double
fabric model (SAN A/SAN B). The separation of the two SANs is logically
implemented as two different VSANs that map to two different VLANs (VLAN A and
B). Fibre channel traffic in SAN A becomes the FCoE traffic in VLAN A, the
Fiber Channel traffic in SAN B becomes the FCoE traffic in VLAN B, and the LAN
traffic is carried on one or more additional VLANs over the converged Ethernet
infrastructure. In this logical environment, the VSAN A/VSAN B configuration
protects against fabric-wide control plane failures.
The traditional method
of hosts that connect to two separate SANs is still supported with the FCoE
over FabricPath architecture. The host is connected to two different leaf nodes
that host a disjointed set of VSANs. Beyond these leaf nodes, the fabric is
converged on the same infrastructure, but the host continues to see two SAN
The following figure
shows a FabricPath topology with n spines (S) and m leafs (L). The m leafs
communicate to each other through the n spines using FabricPath encapsulation.
Figure 1. FabricPath
FCoE creates an
overlay of FCoE virtual links on top of the underlying Ethernet topology,
irrespective of how that Ethernet topology is constructed and which protocol is
used to compute the MAC address routes.
In a dynamic FCoE
environment, the topology is developed using the leafs as FCoE Forwarder (FCF)
switches that are forwarded through transparent spines.
FCoE hosts and FCoE
storage devices are connected to a FabricPath topology through the leaf
switches. In this configuration, only the leaf switches perform FCoE forwarding
(only the leaf switches behave as FCFs); the spine switches just forward
MAC-in-MAC encapsulated Ethernet frames that are based on the outer destination
The following figure
shows the logical FCoE overlay topology of VE_Port to VE_Port virtual links on
a FabricPath topology.
Figure 2. FCoE Overlay of
VE_Port to VE_Port Virtual Links
Only the FCFs, that
are implemented by the leaf switches are part of this overlay topology. This
topology is seen by Fabric Shortest Path First (FSPF), for each FCoE VLAN. FSPF
computes over which virtual link to forward an FCoE frame based on its DomainID
(D_ID). A virtual link is uniquely identified by the pair of MAC addresses
associated with the two VE_Ports logically connected by it. Identifying the
virtual link is equivalent to identifying which MAC addresses to use for the
FCoE encapsulation on the transport network.
Use Lm as the number of
leafs that are feature enabled. The feature might not be enabled on all leafs.
The FCoE mesh is basically the leafs where FCoE or FabricPath is enabled.