Server Cluster

Clustering is the ability of Windows Server 2003 to create a server-based application that is scalable and reliable by regulating the balance of client data traffic across multiple servers and provide a backup server when another server is damaged

Technology of Clustering 
In windows server 2003 there are two kinds of clulsters technology, Server Cluster and Network Load Balancing Clusters. the difference lies in the type of application you run, and the characteristics of the data used.

  1. Server Cluster
    Server clusters are designed for the long-running applications on the memory state or the application for which data are large and frequently changing, commonly known as statefull applications and contains a database server such as Microsoft SQL, Exchange servers, file servers and printers. All nodes in this cluster are connected by a set of data shared SCSI bus or SAN - storage area network. all nodes have access to the same application data, and each node can process the client request at any time. You can configure each node of the cluster is to be passive or active. Active nodes that can receive and process requests from clients, while the passive node is idle and serves as a "fallback" if the active node fails.
    This is exemplified in the following figure there are two servers that form a cluster in which both are running Windows Server 2003 and Microsoft SQL server, each server connected to the same NAS device that contains the database. Both servers also have a special connection that is used to detect heart rate respectively in case of malfunction.


    Server A serves as an active node while server B serves as a passive node. Server A is functioning normally all the time, running a database application, receiving requests from the clients database, and accessing the database files on NAS devices. However, but suddenly server A for some reason fails to function, then the server B as the passive node detects a failure of the active server function and immediately became an Active node replaces the function of the server A, processes the request clients use the same database on the NAS.

    • Just like NLB, server cluster also has a separate name and IP address with the IP address of each node. When the active node fails, no change in perceived function on clients, because the system handled directly by the passive node or if the X-Node Clustering passive node is a node survival.
    • Server clusters can only run on a system windows server 2003 Enterprise and Datacenter editions only, for Windows 2003 Standard edition can only run NLB. Each node may not use a different windows 2003 edition, must be uniform, Enterprise or Datacenter editions only.
    • The maximum can be up to 8 nodes that could serve each failover and failback. Failback is not configured by default, so it must be configured manually or automatically. Most practitioners do manual configuration for failback to be able to evaluate the function of the node failure.
    • Requires special hardware disk drives eg Fibre Channel, Shared SCSI or SAN. Fibre Channel is a high-speed serial network technology up to 100 Mbytes per second using full-duplex communication. While SCSI uses parallel signal technology.
    • Used in SQL Server, MS Exchange, File and Print servers etc..

    X-Node Clustering
    Significant development in the system cluster in windows server 2003 is the number of nodes that can be up to 8 nodes for failover and failback. X node clustering gives us the option to create a minimum of 2 nodes with a maximum up to 8 node failover / failback.

  2. Network Load Balancing (NLB) Cluster

    Network Load Balancing in Windows 2003 is a type of clustering is more easily installed, configured, and maintain. You can use existing hardware and applications on your computer and no longer need additional software to install. You can use the application in windows 2003 it is "Network load balancing manager" to make, manage, and monitoring the NLB cluster. NLB's run more applications stateles memory, usually in the form of data that are not frequently changed.

    • Supported by all platforms of Windows Server 2003 Standard, Enterprise and Datacenter.
    • Can use up to 32 server nodes of load balancing, where each other has a duplicate copy of the application that you want to give to the clients.
    • Full load balancing for TCP and UDP traffic
    • Can be used for various types of servers such as WEB, ISA, VPN, Media Server, and Terminal Server


0 comments:

Post a Comment

Write your constructive comment here...

About Me

Hi, I’m Yani and my Syszoom blog is to share everything about the latest information around the technology field. I started Syszoom blog in 2011 to help other people solve their problem about computer, software, etc. I spent over five years as a teacher, and Syszoom blog is implementation of my knowledge in information technology field. To contact me, please email me at ditsucces@gmail.com.
Powered by Blogger.