ENT's Guide to Microsoft Clustering Technologies
Originally posted in May 2001. Last updated in August 2002.
- By Scott Bekker
Over the last several years, Microsoft has greatly expanded its list of technologies that fit under the umbrella term "clustering." Here's an overview of some of the most prominent Microsoft clustering technologies.
Microsoft Cluster Services – Failover clustering for high availability. In this
scenario one production server is connected to another standby server which assumes the workload
if the first server dies. The second server must start the designated application. While the
servers share storage, any in-memory transactions that were ongoing when the first server went
down will be lost. This technology has a two-node limit in Windows NT Server 4.0, Enterprise
Edition and Windows 2000 Advanced Server. Windows 2000 Datacenter Server allows for four-node
clusters, making it possible to have the more cost-effective scenario of one standby server for
three production servers. Known as Microsoft Cluster Service (MSCS), it was formerly a distinct
product called Microsoft Cluster Server and is sometimes still referred to by its code name of
Wolfpack. In Windows .NET Server, Microsoft will bump up failover clustering to support eight-node clusters in both Windows .NET Enterprise Server (the Windows .NET generation version of Windows 2000 Advanced Server) and Windows .NET Datacenter Server.
Network Load Balancing – IP load balancing technology for distributing Web traffic across a farm of servers. Currently supports 32 server nodes. The technology was originally introduced as a Windows NT 4.0 add-on under the name Windows Load Balancing Services (WLBS). The name was changed to Network Load Balancing (NLB) in Windows 2000. Ships as part of Windows 2000
Advanced Server and Windows 2000 Datacenter Server.
Component Load Balancing – Application load balancing for distributing applications to improve performance. CLB distributes application components across multiple servers, which run the components simultaneously to improve the performance of the application. Originally planned as a part of COM+ in Windows 2000, Microsoft decided not to ship the technology standard with Windows 2000. The company made the technology available for free download as a technology preview, then included it in Application Center 2000.
Distributed Partition Views – Database scaling technology that allows database architects to spread a database across several server nodes. In the typical example, a large database would be split so that records A-F would be stored on the first node, G-L on the second
node, and so on. The approach allows Microsoft to overcome the limitation of its traditional cap of eight-processor servers by ganging several servers together for greater scalability. Distributed Partition Views (DPVs) debuted with SQL Server 2000.
System Area Networks – As much a networking technology as a clustering technology, System Area Networks (SAN) allow traffic to bypass the TCP/IP stack for quicker transmission within a secure data-center environment. The term has been around longer than the more common SAN association – Storage Area Network. Microsoft's first support for System Area Networks arrived last fall with Windows 2000 Datacenter Server.
Scott Bekker is editor in chief of Redmond Channel Partner magazine.