THIS IS A TEST INSTANCE. ALL YOUR CHANGES WILL BE LOST!!!!
Introduction
QMF (Qpid Management Framework) is a set of APIs, in multiple programming languages, that provide an abstraction for object-oriented and service-oriented remote management. It is layered on top of AMQP messaging and implemented using the Apache Qpid messaging APIs.
Architectural Components and Operations
There are two central components in the architecture of QMF: The Agent and the Console. An agent is a software component that is managed via QMF and a console is a component that manages agents. The distinction will become clearer as we discuss operations.
A better way to describe console and agent is to call them roles. For example, a software program may perform the role of QMF Agent, and therefore become manageable via QMF. Another program may act in the role of QMF Console and manage some number of agents. It is also possible for a software program to act in both roles simultaneously and can do so using a single connection into the AMQP messaging infrastructure.
+-----------+ +-----------+
| Console | | Console |
+-----+-----+ +-----+-----+
| |
| |
+--------+----------------+-------------------------+
| |
| Standard AMQP Messaging |
| |
+----------+--------------+--------------+----------+
| | |
| | |
+----+----+ +----+----+ +----+----+
| Agent | | Agent | | Agent |
+---------+ +---------+ +---------+
In QMF, there may be any number of agents and any number of consoles. Of course, if there are no agents, it won't be a very interesting network. Agents and consoles may appear and disappear dynamically.
Operations
QMF operations are defined in terms of the console and agent roles. This section will introduce the operations at a fairly high level. For full details on how the operations operate, refer to the API specification for the programming language you are interested in.
Agent Locate
After a console first connects itself to an AMQP messaging broker, the first thing it must do is identify one or more agents with which to interact. This is done in one of two ways, depending on the purpose of the console application.
The console may attempt to locate a single specific agent by name (i.e. The HP printer on the second floor). In this case, the console communicates directly with the agent to see if it is present in the network (using a direct exchange and the agent's unique name).
Alternatively, the console may wish to interact with a set of agents, possibly all of them. In this case, the console establishes a set of criteria in the form of a query (i.e. "all printers where vendor == 'HP'") and builds a list of available agents. In this case, the console sends a multicast query (using a topic exchange) and collects responses from matching agents. Furthermore, if a new agent comes on line that meets the criteria, the console will learn about the new agent.