Administrator Guide

CAUTION: For iSCSI only, servers initiate I/O to iSCSI ports through the control port of the fault domain. If an iSCSI
port moves to a different fault domain, its control port changes. This change disrupts any service initiated through
the previous control port. If an iSCSI port moves to a different fault domain, you must reconfigure the server-side
iSCSI initiators before service can be resumed.
• For each fault domain, it is a best practice to connect at least two cables from each controller to the Fibre Channel fabric or Ethernet
network.
Fault Domains in Legacy Mode
In Legacy Mode, each pair of primary and reserved ports are grouped into a fault domain. The fault domain determines which ports are
allowed to fail over to each other.
The following requirements apply to fault domains in legacy mode on a dual-controller Storage Center:
• A fault domain must contain one type of transport media (FC or iSCSI, but not both).
• A fault domain must contain one primary port and one reserved port.
• The reserved port must be located on a different controller than the primary port.
NOTE: For a single-controller Storage Center, only one fault domain is required for each transport type (FC or iSCSI)
because there are no reserved ports.
Failover Behavior
Legacy mode, ALUA port mode, and virtual port mode behave differently during failure conditions because they use different mechanisms
to provide fault tolerance.
Table 12. Front-End I/O Ports Failover Behavior
Scenario Virtual Port Mode Legacy Mode ALUA Port Mode
Normal operating
conditions
All ports are active and pass I/O.
• Primary ports pass I/O.
• Reserved ports remain in a
standby mode until a controller
failure.
• Active/Optimized ports pass
I/O.
• Standby ports remain in a
standby mode until a controller
or port failure.
A controller fails in a dual-
controller Storage Center
Virtual ports on the failed
controller move to physical ports
on the functioning controller.
Primary ports on the failed
controller fail over to reserved
ports on the functioning
controller.
Active/Optimized ports on the
failed controller fail over to the
Standby ports on the functioning
controller.
A single port fails (single- or
dual-controller Storage
Center)
1
An individual port fails over to
another port on the same
controller in the same fault
domain.
The port does not fail over
because there was no controller
failure. If a second path is
available, MPIO software on the
server provides fault tolerance.
The port fails over to the Standby
port on the functioning controller.
NOTE:
1
To support port level failover, a controller must have at least two ports in the same fault domain using the same
transport media, such as FC or iSCSI.
Rebalancing Front-End Ports
If a controller has been added or taken offline, ports can become unbalanced. If local ports are unbalanced, you are prompted to balance
the ports by a message at the top of the Summary tab.
About this task
NOTE:
Front-end ports are automatically rebalanced when using SCv2000 and SCv3000 series controllers. Manual port
rebalance is not necessary.
Steps
1. If the Storage Manager Client is connected to a Data Collector, select a Storage Center from the Storage view.
Storage Center Maintenance
243