High Availability; Failover Packages - HP StorageWorks 8000 - NAS Overview

High Availability

8 Network and Storage Planning
The HP NAS 8000 HA solution includes a
a unified system. The cluster is managed by a Quorum server.
The main purpose of high-availability clusters is to provide a higher degree of storage
availability to client systems than is possible with a single server. This is accomplished by
eliminating single points of failure and providing functional redundancy. Uninterrupted service
is provided by failing over file serving capabilities (see "Failover Packages" on page 8) to a
secondary server in the event of a failure in the primary server in the cluster.
The NAS servers provide failover services for each other.
I
The Quorum server manages the NAS servers, or
I
arbitration services that prevent data corruption.
Clusters
HA Clusters provide a higher degree of storage availability than is possible with a single
server. This is provided by:
Hardware component redundancy designed to eliminate single points of failure. Each
I
cluster consists of two or more NAS servers (nodes) plus one Quorum server.
Software failover capabilities. All storage is controlled by packages which are the smallest
I
units of failover in the cluster.

Failover Packages

When a package failure is detected, the package is halted and will fail over to another node,
provided that an eligible adoptive node is present. Each package and its associated resources
are monitored independently. This allows failures that are limited to a single package to be
handled without affecting the state of other packages.
The NAS solution supports two failover models:
Active/Active (or "Dual Active") means that all cluster nodes can be serving data
I
simultaneously. The nodes will provide failover services for each other, but can be fully
utilized at all times. This mode is achieved by configuring failover packages to run on both
cluster nodes – effectively dividing the storage between the nodes.
Active/Passive (or "Single Active") means that one server is actively serving data while the
I
other node is in full-time stand-by mode, becoming active only when the other node fails.
This mode can be realized by configuring all of the failover packages and storage to run on
only one node. Without any packages to run, the other node will have nothing to do but
wait and monitor the other node for failure.
cluster , or group, of two NAS servers that function as
cluster nodes , and provides cluster