Windows Server cluster technology (secret) technology revealed

Windows Server cluster technology is designed as a three-tier application architecture, and Microsoft architects have implemented solutions for each layer. Next, I introduced the Network Load Balance (NLB) to the front-end clustering technology. Others include the COM+ component load balancing (CLB) of the mediation layer and the cluster belonging to the back-end server (Microsoft). Cluster Server, MSCS). Before introducing NLB, let's take a look at cluster-related knowledge: the definition of clusters, and how clusters can help us. The basic concept of clustering Simply put, cluster refers to a group of servers that work together to provide the same service or application to improve service or application operability (Note 1) and reliability ( Reliability, note 2), and scalability (Note 3). When a server in the cluster hangs, the request is transferred to other servers, providing a year-round, uninterrupted service. Windows Server supports three cluster technologies: NLB, CLB, and MSCS. NLB and MSCS are integrated into Windows Server, while CLB requires the purchase of Application Center. These clustering technologies are used as follows: 1. NLB provides network traffic load balancing for TCP/IP-based services and applications to improve operating system availability and scalability. Common applications include terminal services, web services, and web applications. NLB provides external services through a virtual IP. When a request is received, NLB randomly determines which server will handle the request. 2.CLB, which provides dynamic load balancing of mediation applications that use COM+ components to increase system availability and scalability. CLB will decide who will process the service request based on the current workload. 3. MSCS, providing fault-tolerant transfer of back-end services and applications to improve system availability. Common applications are SQL Server and Exchange Server. MSCS is determined by the client to handle the service request. All servers share a share storage to store the session state. When the active server hangs, it continues to be taken over by the passive server. The passive server will take the session state from the share storage and continue the unfinished work to achieve the purpose of fault-tolerant transfer.