I haveing Been working on the cisco UCS systems, More Understanding of Single line of failure is that the redundancy always apply a major response to any hardware failover but if have come across a Scenario , like for instance ,what will happen if FI-A (Fiber interconnect A) and CISCO 9148 MDS-B( SAN Switch) fail or vice versa , How will the Uplink port responses to blade server to Northbound network and if there is no physical connection b/n them and assuming logical connection exists?
The Best Practice for setting up UCS is to uplink to Cluster-base (vPC or VSS) Switches for Ethernet, configured in a classic bow-tie cabling, and to configure FI-A directly to MDS-A, and FI-B directly to FI-B. On the FC side of things, the cross-cabling that supports multi-pathing occurs between the Fabric Switches and the SA Controllers.
For Ethernet, Fabric HA is achieved either with UCS Fabric Failover (Ethernet Only), or OS bonding/teaming drivers from within the OS.
For FC, Fabric HA is achieved through 3rd party multi-pathing software, and the UCS Service Profile providing vHBA's connected to the A Fabric and the B Fabric. Typically, there are at least 2 vHBA's deployed in a Service Profile....vhba-A, and vhba-B.
Typically, the logical Service Profile Configuration has connectivity to both the A Fabric and the B Fabric for vNICs and vHBAs....
If you've set-up you environment in this way, you can actually power-down either FI-A or FI-B and still maintain Ethernet/FC connectivity. This is a common test for Cisco Advanced Services at the completion of an engagement.
It's also important to remember, in UCS, the Blades are always cabled internally, through the chassis mid-plane to BOTH Fabrics, A and B. Blades are always pinned with KR traces to each IOM, A and B. Whether or not you choose to leverage this is a function of the logical network and storage configuration of the vNICs and vHBAs within the Service Profile.
Hope this helps...