This paper explains uplink strategies for traffic coming in and out of a Cisco Unified Computing Solution (UCS) chassis. An uplink can be Ethernet, Fibre Channel, or FCoE and is a physical connection on the FI that leaves the UCS domain directed away from the B Chassis.
In 2014, Cisco Unified Computing Solution (UCS) became the number one server in the US in the x86 blade server market with more than 42 percent market share up from 2.4 percent in 2009. One of the main differentiators in Cisco's server solution is the capacity for two totally redundant failover paths for all server Ethernet and Fibre Channel traffic for every server. The other main differentiator is the Service Profile. See my other white paper titled Solving Issues on creating a UCS Service Profile for more info on troubleshooting Service Profiles.
To illustrate the hardware options in UCS, log in to UCS Manager and under the Equipment tab and click on the Equipment icon. In the right pane, click on the Main Topology View.
Cisco allows an option of either one or two Fabric Interconnects (FI) in a UCS management domain. If you have only one FI, then you will not be able to implement any UCS failover options and the topics in this white paper would not apply. Note that one of the FIs is labeled A, and the other B. These are the "official names" of the FIs and a maximum of two FIs are supported per a UCS management domain.
Note that the above example shows only one chassis, however the concepts in this white paper apply from any of the one through twenty chassis supported by the currently available UCS Manager (UCSM) software versions.
The focus of this white paper is on uplink strategies for traffic coming in and out of a UCS chassis. An uplink can be Ethernet, Fibre Channel, or FCoE and is a physical connection on the FI that leaves the UCS domain directed away from the B Chassis.
Service Profile Configurations
Cisco's implementation of stateless computing allows all the hardware settings of a server to be saved on a location outside the server itself. Stateless computing in UCS is implemented with an XML-based software feature called a Service Profile. In the event of disaster recovery, the amount of time to restore the server to its identical former condition is dramatically reduced with Service Profiles. Service Profiles contain virtually all possible hardware settings of a server, which can be applied, modified, or moved to a different blade.
Service Profile Fundamentals
For starters, if a Service Profile is not associated with a blade (individual hardware server), the blade literally can't even be turned on. After you log in to UCS Manager (UCSM), highlight your server under the Equipment tab.