N1 Service Provisioning System 4.1 User's Guide

Approaches to Modeling

Modeling creates a representation of an application, so that the application can be installed, configured, and managed on targeted host computers. The model for the application can be based on the configuration of a specific host computer running an application, or it can be based on components stored in a repository such as a source code control system.

A host computer that is designated to serve as a reference system for a particular application is called a gold server. [For an overview of the design philosophy behind gold servers, see Traugott and Huddleson, “Bootstrapping an Infrastructure,” Twelfth USENIX Systems Administration Conference (LISA "98), Boston, Massachusetts. This paper, along with others devoted to the issues of IT infrastructure, is available at http://www.infrastructures.org. ] In many organizations, a development team, a QA team, or a data center team creates a gold server, so that IT operators have a tested and approved example of an application that needs to be deployed in a production environment.

Using the N1 Service Provisioning System software , you can model an application from a gold server or some other repository. The component model that you create can:

Creating models makes applications more manageable. An application model can includes vast amounts of information (files, directories, and so on) that it would be difficult or impossible to manage manually.

A Typical J2EE Modeling Process

In J2EE applications, components typically comprises several resources, such as directories, archives, and so on. In modeling a J2EE application, then, you check in the resources that make up a component and then build the component model based on the resources.

These are the steps typically involved in modeling J2EE applications.

  1. Define and configure hosts.

  2. Check in resources that make up each component.

    • Install a Remote Agent on the gold server or development system with the components you want to model.

    • Select the resources you want to include in components.

  3. Define components.

    • In addition to defining the resources that make up a component, you can create variables for configuration parameters, add information about component dependencies, etc.

  4. Write plans for deploying and configuring components.

  5. Run plans.

  6. Run comparisons as necessary to analyze your application environment.

A Typical Windows Modeling Process

Windows application components are usually managed as a whole units, rather than as a collections of distinct resources.

These are the steps typically involved in modeling Windows applications.

  1. Define and configure hosts.

  2. Check in components.

    • Install a Remote Agent on the gold server or development system with the components you want to model.

    • Select the components you want to model. The N1 Service Provisioning System software includes model templates for the most common Windows application components (see Table 1–2 for a list). In many cases, once you have selected the resource template for the Windows component, no further modeling will be necessary.

  3. If appropriate, add additional resources to components.

    • In addition to defining the resources that make up a component, you can create variables for configuration parameters, add information about component dependencies, etc.

  4. Write plans for deploying and configuring components.

  5. Run plans.

  6. Run comparisons as necessary to analyze your application environment.