Browse Prior Art Database

Dynamic targeting of resources in a pure adaptive environment

IP.com Disclosure Number: IPCOM000030168D
Original Publication Date: 2004-Jul-30
Included in the Prior Art Database: 2004-Jul-30
Document File: 3 page(s) / 88K

Publishing Venue

IBM

Abstract

The methodology provides an approach to dynamically target resources of IT environment that is going to be managed leveraging on an adaptive management paradigm. The term "dynamically" means that as new resources come up they automatically get managed.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

Dynamic targeting of resources in a pure adaptive environment

Adaptive management paradigm requires IT Administrators to put under form of rules the management of the IT environment they want to put in place. A rule tyipically consists of a targeting part, which identifies the set of managed targets receiving the rule, and a configuration element that requires a managed resource to be in a well-defined desired state. Targets get the rules according to their current state (intended as the set of characteristics of the resources hosted on that target) and perform some management actions to bring the specified resources into the desired state.

    The problem that is addressed in this paper concerns the language to be adopted for targeting the rules. Simply, a boolean expression made up by key-value pairs may be used. Nevertheless, only a subset of potential targeting conditions can be expressed by means of such a formalism. I.e. let's think about a situation in which the rule builder wants to deliver a rule just to those targets where an instance of a specific application interfaces a well-known version of a database: "all the targets where my_application is running and uses Oracle at version 8". Unfortunately, the key-value pairs formalism is not enough expressive to allow such a condition to be expressed due to the fact that multiple instances of the same resource type cannot be described.

    Let's introduce a concept of grouping of key-value pairs. The state of each target can hence be described by a set of groups of key-value pairs, one for each resource instance in the scope of that target. Relationships between instances are being modeled using a key binding the two instances of the relation. In fact, to model for example the fact that an application uses a database, application instance's group may have a key whose value is equal to the one of a key from the database instance's group.

    The algorithm advocated to calculate the rules for each target allows selecting from the target's state those resource instances that satisfy the targeting expression of the rule. Those instances can be referenced in the configuration part of the rule in case managed resources are part of the target's state and have to be dynamically selected according to a specific condition. In our example, it could be required to configure only those instances of the application that use Oracle at version 8.

    Let's see how the algoritm works using an example. Suppose the following targets present to the rule repository with their state:

TARGET 1

[GROUP]

     [DatabaseVendor=SQL] [DatabaseInstance=D1] [Datab...