Network topology analysis utilising the Activity Service to send informative context across a distributed server environment
Original Publication Date: 2001-Dec-16
Included in the Prior Art Database: 2003-Jun-20
Disclosed is a mechanism for providing lightweight network topology and performance analysis, to transport dynamic routing and performance statistics through a distributed server environment. The Object Management Group (OMG) Activity service is a Common Object Request Broker Architecture (CORBA*) Object Service described in the OMG document "Additional Structuring Mechanisms for the OTS" by IBM** et al, 2000. The J2EE Activity Service is based on the OMG Activity Service and is described in the Java Community Process specification JSR095. These services provide a generalized framework to support applications that require non-prescriptive, extended transaction behaviour. This is achieved by providing a generalized context management service on which a High Level Service (HLS) such as the Object Transaction Service (OTS) can be built. The HLS sits between the Activity Service and an application and provides services to the application in support of the specific extended transaction model. The functionality of the HLS is implemented in part by the HLS itself and in part by the Activity Service through delegation from the HLS. This architecture is described in the documents referenced above. In a distributed server environment, remote method requests flow between nodes in a network of application servers, usually originating from a given entry point such as a web server. The execution path of the requests through the network may be non-deterministic. For example, some form of workload management may be in operation and the result is that the exact route taken by a request is not known. Other statistics such as the performance of these requests or the performance of the individual links that make up the network may also be unknown. The disclosed mechanism constructs a new HLS using the Activity Service framework. The purpose of this new HLS is to gather such statistical information as outlined below. The exact statistics collected could vary between implementations, for example: the remote method execution path through the network; the duration of a remote method possibly broken down on a node by node basis; the node connection types encountered; network traffic level information; network failure information.