Disaster Recovery in Windows 2003 – Clustering

Introduction to Windows Server 2003 Clustering

Clustering is one of those computing concepts that is – simply great. Also in my ‘simply great’ category are: client server, DHCP, and VPN.  These are technologies that just make sense and should be implemented where ever possible.  The twin benefits of clustering are load balancing and fault tolerance.  One day soon SAN (Storage Area Network) and wireless networks will join clustering in that ‘simply great’ category.

There are at least two distinct clustering strategies: Stateful and Stateless.  Stateful is proper fault tolerance clustering, whereas Stateless is just load balancing.  The key points to check are, what application the clustering is designed for?  Does it need special hardware?

Stateless NLB Clusters (Network Load Balancing)

Examples: VPN, IIS or Terminal Server

This is the clustering that you configured through the Network Adapter Properties.

It supports up to 32 nodes, is Stateless and automatically balances the load when more hosts come on line. To further administer NLB there is a little known utility nlbmgr which you can launch from the run command.

A feature of all clustering types is that each server needs a second network adapter (dedicated IP address) which is just used to communicate with the other cluster servers.

From the host point of view they connect to a cluster, so each cluster needs a DNS name and a corresponding cluster IP address.

Which ever method of clustering you use, make sure you purchase the Enterprise or Datacenter edition of Windows 2003.  The standard edition does not support clustering.

Stateful Server Clusters

Examples:  SQL or Exchange

This method of clustering is controlled by a service that you install through Add or Remove Programs.  Note: Cluster Service requires an account much like Exchange 5.5.

Server clusters support up to 8 nodes and are Stateful.  This type of clustering supports failover and is designed primarily for fault tolerance. 

The second network card maintains a ‘heartbeat’ with other members of the cluster.  The virtual cluster name needs to be registered in DNS.

To administer your cluster, Start (menu) Run, CluAdmin.exe.  The concept is that one node owns the resource, and the other nodes are aware and take over in the event of failure.

Applications such as SQL and Exchange have their own administration programs to configure Active / Active or Active / Passive clustering.  Where possible favour Active / Passive clustering, this means that in the event of failure all the functions are taken over by a ‘hot spare’ server.

Guy Recommends:  A Free Trial of the Network Performance Monitor (NPM)Review of Orion NPM v12 v12

SolarWinds’ Network Performance Monitor will help you discover what’s happening on your network.  This utility will also guide you through troubleshooting; the dashboard will indicate whether the root cause is a broken link, faulty equipment or resource overload.

Perhaps the NPM’s best feature is the way it suggests solutions to network problems.  Its second best feature is the ability to monitor the health of individual VMware virtual machines.  If you are interested in troubleshooting, and creating network maps, then I recommend that you give this Network Performance Monitor a try.

Download your free trial of SolarWinds Network Performance Monitor.

Limitations of Disk configuration on server clusters.

Remember that the disks are likely to be the area where you need specialist cluster aware hardware.  At the top of the range, the storage will be handled by NAS and SAN systems.

Disks to be used for cluster storage must be configured as basic disks and must be formatted as NTFS. You cannot configure cluster storage devices as dynamic disks or spanned volumes (volume sets) if they will be used as cluster resources.  You can, however, use the DiskPart.exe utility to extend the volume of a basic cluster disk. for more information, see DiskPart.

It is highly recommended that you do not enable write caching on cluster disks.
In server clusters, the process of taking resource groups offline on one node and bringing them online on another node. When failover occurs, all resources within a resource group fail over in a predefined order; resources that depend on other resources are taken offline before, and are brought back online after, the resources on which they depend.

If you like this page then please share it with your friends


More Topics for Disaster Recovery in Server 2008