An Application Center cluster is a set of servers that serve the same content to clients (for example, Web pages and COM+ components). Application Center simplifies the creation and management of clusters through the use of wizards and a Microsoft Management Console (MMC)–based user interface.
Application Center supports the following cluster types:
General (Web-tier) cluster
COM+ application cluster
COM+ routing cluster
General (Web-tier) Cluster
A general cluster is any standard cluster that uses a group of servers to process client requests. These clusters can include:
Stand-alone servers (clusters with one member only)
COM+ Application Cluster
A COM+ application cluster handles processing of COM+ components exclusively. When Web servers or clusters instantiate objects for COM+ components, the method calls to instantiate the objects can be load balanced across multiple cluster members within a COM+ cluster.
For clusters that handle both COM+ activations and Web sites, use the general cluster type.
COM+ Routing Cluster
A COM+ routing cluster uses Component Load Balancing (CLB) to route requests for COM+ components from a general cluster to a COM+ application cluster.
Note Most of the time, COM+ routing clusters are not needed. Web clusters can communicate to back-end COM+ clusters over load-balanced DCOM without an intervening COM+ routing cluster. This works because each general cluster member acts as its own router for determining which remote COM+ server should receive remote COM+ component activation.
COM+ routing clusters are useful when COM+ load balancing is required from one general cluster across multiple COM+ application clusters. This is because this release of Application Center CLB is limited to a single list of remote servers for every routing server (that is, every member in a general cluster). This means that if you require load-balanced COM+ activations from one general cluster across multiple COM+ clusters, a COM+ routing cluster is required for load balancing the second and subsequent COM+ clusters.
A COM+ routing cluster typically comprises two members that are using Network Load Balancing (NLB) to ensure availability of the COM+ router.
Top of page
Load balancing is used to distribute workload among cluster members. Application Center supports three types of load balancing:
Other load-balancing devices
NLB is the IP load-balancing technology that is provided as part of Microsoft Windows® 2000 Advanced Server and Microsoft Windows 2000 Datacenter Server. It is also shipped with Application Center to allow Windows 2000 Server to support NLB while running Application Center. With NLB, incoming TCP traffic, User Datagram Protocol (UDP) traffic, and Generic Routing Encapsulation (GRE) traffic requests are distributed among cluster members.
COM+ allows developers to write software components in a variety of languages. Because the location of COM+ components is transparent, they can run in a distributed environment. Using CLB, Application Center takes advantage of this to provide load-balanced activation of COM+ components across a cluster. CLB ships as part of Application Center.
Application Center is compatible with other load-balancing devices and they can be used together with the deployment and monitoring capabilities provided by Application Center. However, in this version, Application Center does not actively manage these devices. The Microsoft Application Center Resource Kit includes software that allows limited integration of other load-balancing devices. This software allows you to set members offline and online from both the Application Center snap-in and the command line. You can also use a Windows Management Interface (WMI) provider (included in the Resource Kit) for the device to monitor the state of cluster members.
Top of page
Synchronization and Deployment
Application Center manages cluster synchronization and deployment. Application Center uses a single application image to represent all of the required content for a solution (for example, a Web site). The application image can include Web sites, Web pages, COM+ applications, system settings, and any other content that must be installed on each cluster member.
Deployment is the process of copying an initial application image from one cluster to another. Typically, deployment occurs from a staging, or a prepatory, cluster to a production cluster that holds the live content. During deployment, the application image is copied to the cluster controller of the recipient cluster.
A cluster controller maintains the latest, most accurate content for a cluster. The controller is responsible for synchronizing its content to all of the cluster members. If the content on the controller changes, perhaps as a result of an incoming deployment, changes are synchronized automatically to each cluster member. You can also perform synchronization manually.
Note Application Center is not intended for use as a deployment tool for operating systems, system components (such as Service Packs [SPs]), or computer programs, such as Microsoft Word.
Top of page
To maximize availability, you must understand events and performance metrics of a cluster. For example, a certain event might represent impending failure, or a CPU utilization metric creeping upwards of 90 percent can indicate that you need to add an additional server to the cluster. Application Center provides a rich set of monitoring features that allow you to detect, analyze, and fix performance and availability problems.
Events and performance metrics are stored on each member. Performance metrics can be aggregated across a cluster and viewed in real-time or over long periods, enabling you to spot immediate problems or chart long-running trends. Application Center also rolls up events from multiple members into a single view, which expedites processing and filtering events for the system administrator.
The Microsoft Health Monitor 2.1 functionality that is built into Application Center allows you to set thresholds on a wide variety of system and application information, such as Windows Event Logs, HTTP calls, Windows service failures, any WMI event or instance, and when thresholds are exceeded. You can use this functionality to send automated e-mail notification or run command-line scripts.
Top of page
The primary administrative interface for Application Center is the MMC, which is a Windows 2000 tool that is used to create, save, and open collections of administrative tools, called consoles.
Through MMC, you can administer an Application Center cluster locally or through a secure remote connection. Additionally, limited administration can be performed remotely via the Web-based management interface (the Web-based Administrative client).
Whichever user interface is used, Application Center provides a view of the cluster and its members. With both the Application Center MMC snap-in and Web-based Administrative client, you can administer more than one cluster at the same time.
Application Center also includes a command-line tool, which you can use to automate administration and management of Application Center.