Microsoft Cluster Server

From Wikipedia, the free encyclopedia

Microsoft Cluster Server (MSCS) is software designed to allow servers to work together as one machine, to provide failover and increased availability of applications, or parallel calculating power in case of high-performance computing (HPC) clusters (as in supercomputing).

1 Background
2 Failover Clustering
3 High-performance computing (HPC) clustering
4 Cluster Configurations
5 Efficiency
6 See also
7 References
8 External links
- 8.1 News

[edit] Background

Windows NT Server, Enterprise Edition was the first version of Windows server to include the MSCS software. The software has since been updated with each new Enterprise Edition server release: Windows 2000 Advanced Server, Windows 2000 Datacenter Server,Windows Server 2003 Enterprise Edition And will be presented with many new features in the upcoming Windows Longhorn Server. The cluster software evaluates the resources of servers in the cluster and chooses which are used based on criteria set in the administration module. In June 2006 Microsoft also released Windows Compute Cluster Server 2003,^[1] the first high-performance computing (HPC) cluster technology offering from Microsoft.

[edit] Failover Clustering

Unlike other clusters which are made for better performance, Windows clusters currently present the option of failover services only, which means that if a failure occurs on a server that is a member of the cluster (Cluster node) the services that the failing server was hosting will automatically restart themselves on another server that is a member of the same cluster. The process of a service moving from one server to another is called Failover.

[edit] High-performance computing (HPC) clustering

High-performance computing (HPC) cluster technology is used in supercomputers (multiple computers working in parallel and acting as one computer to achieve very high calculating power). Unlike failover-clusters, HPC-clusters do not required shared disk as it only does calculations and do not need to read or write to a common storage, and if one of the machines in a HPC-cluster fails then all that happens is that the cluster as a whole can not calculate as fast but none of the functionality is lost. Microsoft Windows Compute Cluster Server 2003 (CCS) is a such cluster technology software, announced in June 2006 by Microsoft,^[2]

[edit] Cluster Configurations

Every service or group of services that the cluster runs use resources of the cluster nodes. Every service or a group of services has its own Harddrive assigned to it (which is shared with the other failover cluster nodes), it has its own IP Address and it has its own Network name. All of the resources that a clustered service use are called a Resource Group. The Resource group contains the basic resources that every service need, Disk Drive, IP Address, Network Name, and the service itself. All of those together form a virtual server that can be moved from one server to another in a matter of seconds (Failover) without any dependence in a specific server. The user that accesses this virtual server will be exposed to it like any other server, If it is a File server, Web Server, DHCP/DNS/WINS server, Exchange Server, SQL Server or any other Windows service.

Physical scheme of a windows cluster

[edit] Efficiency

As of today windows cluster servers are not considered very efficient and reliable due to the lack of experience that Microsoft accumulated in the area of clustering compared to other platforms like VMS and Unix platforms that supports clustering from much older versions. The coming windows Longhorn server promises much more upgrades for its cluster services that should make it much more reliable from the previous version of windows.