HP NAS 4000s, Administration Manual

Page: 203 / 246

Cluster Administration
203 NAS 4000s and 9000s Administration Guide
For example, if from a client a network share map F: was established and assigned to
\\Node1\Eng1 instead of \\Fileserver1\Eng1, when Node1 fails and Node2 assumes ownership,
the map will become invalid because the reference in the map is to \\Node1. If the map were
created to the virtual name and Node1 were to fail, the map would still exist when the group
associated with Eng1 failed over to Node2.
The previous diagram is an example and is not intended to imply limitations of a single group
or node. Groups can contain multiple physical disks resources and file shares and nodes can
have multiple groups, as shown by the group owned by Node2.
Cluster Planning
Clustering the NAS 4000s or 9000s greatly enhances the availability of file service by enabling
file shares to fail over to additional NAS devices, if problems arise. Clients see only a brief
interruption of service as the file share resource transitions from one server node to the other.
Requirements for taking advantage of clustering include:
■
Storage planning
■
Network planning
■
Protocol planning
Storage Planning
For clustering, a storage unit (LUN) must be designated for the cluster and configured as a
mirrorset. This LUN is used for the Quorum disk. The Quorum disk is the shared storage used
by the cluster nodes to coordinate the internal cluster state.
One or more RAID arrays are dedicated to each cluster node for data storage. Each cluster
node will assume ownership of at least one physical disk resource. That owner node will serve
all shares within that physical disks resource, until a failover condition occurs. When a failover
occurs, the physical disk resource and all associated shares will transition over to the
remaining nodes and will remain there until the other node is returned to service. Some types
of shares are not cluster aware and will not be available during a failover condition. See the
“Protocol Planning” section for additional information.
To prepare a basic disk for use in a cluster, a cluster group for each basic disk should be
created to allow each resource to failover separately. Once the group is created, a physical disk
resource is created in each of the groups. Cluster groups may contain more than one physical
disk depending on the site-specific requirements. This physical disk resource is required for
basic disk to successfully work in a cluster environment protecting it from simultaneous access
from each node.
Note: The LUN underlying the basic disk should be presented to only one node of the cluster using
selective storage presentation SAN switch zoning or having only one node online at all times until
such times as the physical resource for the basic disk is established.