Table of ContentsView in Frames

System configurations for SnapManager on a Windows cluster using LUNs

System configurations for SnapManager on a Windows cluster

When installing SnapManager on an MSCS cluster, consider the following system configuration requirements and limitations.
Note: No configuration is required if your configuration uses SMB shares only.
Note: You do not need to stop SQL Server instances before or during the SnapManager software installation process.

SnapManager must be installed on all nodes SnapManager must be installed on all nodes of the cluster so that SnapManager backup and restore operations can be performed from any node.

SnapDrive must be installed on a virtual machine running the installer If you are running the installer on a virtual machine using VMDKs, SnapDrive must be installed on that machine.

Maximum cluster size SnapManager supports a maximum cluster size of eight nodes.

Multiple-instance clusters: SnapManager supports multiple-instance clusters, provided that the following additional system requirements are met:

Disk requirements for SnapManager on a Windows cluster

In a clustered environment, SnapManager disk requirements vary, depending on the cluster configuration.

Single-instance cluster example In an active/passive two-node configuration, there are two clustered nodes and one SQL Server instance. If the active node (the node running SQL Server) fails, the cluster transfers the SQL Server instance to the other (previously passive) node, which then becomes the active node and takes over the LUNs previously used by the failed node.

For a single-instance SQL Server cluster, if your SQL Server data is on a shared resource, your disk requirements are the same as for a stand-alone SQL Server system. A LUN gets added for the quorum disk. A minimum of three LUNs are required:

Multiple-instance cluster example In an active/active two-node configuration, there are two clustered nodes and an SQL Server instance running on each node. If one node fails, the other node takes over the SQL Server instance running on the failed node. Because both nodes need to be able to run an active SQL Server instance, each node requires its own disks, as if it were a self-contained, stand-alone system. In addition, one extra LUN is needed for the quorum disk, if a shared quorum disk is used. Whether you use a hard disk or a LUN as the quorum disk, each configuration requires a minimum of five disks used for the following purposes:

Each node must be able to own all clustered disk resources in a cluster at any time.

For more information about MSCS clustering with SQL Server, the SQL Server 2005 Failover Clustering document at the following URL: www.microsoft.com/downloads/details.aspx?FamilyID=818234dc-a17b-4f09-b282-c6830fead499&displaylang=en.