Executive Summary:
Microsoft Windows Server 2008’s failover clustering provides increased hardware support, a single quorum model that can operate in one of four modes, and a simplified cluster creation process.
|
One of Windows Server 2008's advantages is high availability, including failover clustering. Specifically, Server 2008's failover clustering provides increased hardware support, a single quorum model that can operate in one of four modes, and a simplified cluster creation process. (For a basic overview of failover clustering, see the Learning Path.)
Other changes in Server 2008 that improve failover clustering include the configuration flexibility to cater for higher latencies between locations than is typically acceptable for cluster heartbeats, as well as multi-site clustering, which allows a cluster to have IP addresses on multiple subnets and therefore nodes in multiple sites. In addition, Server 2008 failover clustering provides greater scalability because it allows 16-node clusters on a 64-bit architecture, compared with the 8-node maximum in Server 2003 (and Server 2008 32-bit). I'll touch on some of these enhancements as I discuss the main improvements in Server 2008 failover clustering.
Hardware Support
Windows Server 2003 has a dedicated cluster-certified hardware list. For Server 2008, you must purchase logo-certified components, then run a cluster validation process to ensure that your configuration is supported. If your configuration passes the validation test, then your cluster is supported by Microsoft. You can also run the validation process on a configured cluster to check for problems that might be causing the cluster to fail.
The cluster validation process isn't a new tool—it was formerly called ClusPrep and was downloadable from Microsoft.ClusPrep was enhanced in version 2.0, which is included in Server 2008. Four types of tests are included in the cluster validation process: Inventory, Network, Storage, and System Configuration.
Inventory. The Inventory test collects the following information: BIOS information, environment variables, Fibre Channel, Serial Attached SCSI (SAS) host bus adapters (HBAs), iSCSI HBAs, memory, OS, Plug and Play (PnP) devices, running process, services, software updates, system and unsigned drivers, general system information (e.g., computer model, domain).
Network. The Network test checks the cluster network configuration. This check verifies the IP and subnet information on adapters, as well as validates the IP configuration to ensure that addresses are unique, multiple adapters aren't connected to the same subnet, a default gateway is configured on one adapter, and no duplicate MAC addresses exist (which is important if you're using virtual machines—VMs). Network communication between the nodes is checked, and the firewall rules are verified to ensure that cluster communication won't be interrupted.
Storage. The Storage test lists all the disks, particularly shared disks, that are visible to all nodes. Shared disks are then tested for failover suitability, including data being kept intact.
System Configuration. The System Configuration test verifies that Active Directory (AD) is configured correctly and that all the nodes are in the same domain and ideally the same organizational unit (OU), which is important to ensure consistent Group Policy application. Note that the same OU isn't mandatory and will generate only a warning. Nor do nodes need to be in the same subnet or AD site. The System Configuration test also verifies that all the drivers are signed and that the same OS version, service packs, and software updates are installed. Required clustering services (e.g., RpcSs, Remote Registry, LanmanServer, WinMgmt) are checked to ensure they're running. Finally, all the nodes are checked to ensure that they're running the same architecture, which is necessary in a cluster.
The cluster validation process is initiated from the Microsoft Management Console (MMC) Failover Cluster Management snap-in. Select the Validate a Configuration action, enter the names of the nodes that will be part of the cluster, and select the tests to run. Figure 1 shows the cluster validation tests running. When the tests are complete, a summary display shows the status of each test and highlights areas requiring attention or components that aren't suitable for the cluster. For a more detailed report, click the View Report button or go to the the \%windir%\Cluster\Reports folder.
To prevent the problem of buying and installing components that aren't supported, Microsoft offers the Failover Cluster Configuration Program (FCCP). To find cluster configurations that have been previously validated by Microsoft or its partners, go to Microsoft's Windows Server Cluster Solutions website.