Just like switch-level redundancy, fabric-level redundancy has separate serverswitch-storage
paths for data. However, in this case, the switches belong to
separate fabrics. There are no SPOFs. The advantages of fabric-level redundancy
are as follows:
It eliminates the risk of outage if the fabric fails. Human errors such as
improper switch replacement procedures, erroneous fabric configuration
settings, and fabric service failure can take down the fabric.
■■ The number of total available ports is higher. This, of course, comes at
the expense of adding switches.
■■ Two fabrics allow for non disruptive firmware and software code
updates. Services can be failed from Fabric 1 to Fabric 2 (in Figure 23-3);
then you can upgrade firmware in Fabric 1 switches and failback operations
to Fabric 1.
Besides building fabric resiliency, the data in the storage arrays must be protected
against disk failures using several mechanisms. The volumes in the
storage units must be configured as mirrored or RAID-5 volumes. Certain
disks must be designated as hot spares to automatically replace a failed disk in
a mirrored or RAID-5 volume. The disks must be hot-swappable to allow you
to replace failed disks without having to off-line the storage array.