IBM N Series, Hardware Manual

Page: 93 / 368

Chapter 7. Highly Available controller pairs
73
򐂰 Nondisruptive hardware maintenance: When you halt one node and allow takeover, the
partner node continues to serve data for the halted node. You can then replace or repair
hardware in the node you halted.
Figure 7-2 shows an HA pair where Controller A failed and Controller B took over services
from the failing node.
Figure 7-2 Failover configuration
7.1.2 Characteristics of nodes in an HA pair
To configure and manage nodes in an HA pair, you must know the following characteristics
that all types of HA pairs have in common:
򐂰
HA pairs are connected to each other. This connection can be through an HA interconnect
that consists of adapters and cable, or, in systems with two controllers in the same
chassis, through an internal interconnect. The nodes use the interconnect to perform the
following tasks:
– Continually check whether the other node is functioning.
– Mirror log data for each other’s NVRAM.
– Synchronize each other’s time.
򐂰
They use two or more disk shelf loops (or third-party storage) in which the following
conditions apply:
– Each node manages its own disks or array LUNs.
– Each node in takeover mode manages the disks or array LUNs of its partner. For
third-party storage, the partner node takes over read/write access to the array LUNs
that are owned by the failed node until the failed node becomes available again.
Clarification: Disk ownership is established by Data ONTAP or the administrator,
rather than by the disk shelf to which the disk is attached.