Electrical computers and digital processing systems: multicomput – Computer network managing – Computer network monitoring
Reexamination Certificate
1999-08-03
2004-03-30
Cardone, Jason D. (Department: 2142)
Electrical computers and digital processing systems: multicomput
Computer network managing
Computer network monitoring
C709S241000, C714S004110
Reexamination Certificate
active
06714976
ABSTRACT:
FIELD OF THE INVENTION
The invention generally relates to systems and methods for monitoring and managing distributed computing environments, and more particularly, to systems and methods for monitoring enterprise wide operation of a distributed computing system.
BACKGROUND OF THE INVENTION
A distributed computing architecture provides physical and logical distribution computing functions across many computers connected by a network system. Typically, the client initiates a service request to a server across the network. The server responds to the client's request by performing one or more database, file, printing, or other services. During the operation, the client and the server exchange data and individually perform data processing functions necessary for completing the operation. Complexity can arise because a single server can service multiple clients simultaneously, while a client can concurrently access the services of multiple servers. Moreover, servers can act as clients to other servers. Accordingly, distributed computer systems can have complex, multiple tiered distributed architectures.
Despite its complexity, distributed computing architectures have been successful in providing users with sophisticated and powerful systems for efficiently processing large amounts of data and providing rapid digital communication between multiple stations. The power of these systems has lead to the wide-spread proliferation of distributed computing architectures and has further resulted in the development of a plethora of distributed computing services such as client/server databases, distributed applications and networks across heterogeneous environments. Moreover, new technologies continue to fuel the growth of distributed systems. For example, the development of internet and intranet systems suitable for the commercial environment has created a burst of growth in the distributed computing field.
Although distributed computing architectures provide users with efficient and powerful tools, the complexity and sophistication of the architecture make the implementation, deployment, and operation of the actual systems difficult. For example, a typical relational client/server database system will include a database server for providing a number of database services to a plurality of clients. The distributed architecture generally requires that each client is capable of properly communicating with the server, and that the server is capable of coordinating the multiple service requests received from the clients and maintaining data coherency for a data repository that could be distributed among several network memory devices. Loading such a system onto a computer network is a difficult task, made complex because electronic communications occurring between clients elements and servers occurs asynchronously, intermittently and quite rapidly. Accordingly, complex diagnostic and management tools are used to implement these distributed systems and to analyze and improve performance.
The complexity of a distributed computing architecture makes diagnosing system failures and performance analysis a difficult task. The asynchronous and rapid nature of communications between the distributed network components complicates the task significantly. Accordingly, a diagnostic technician may have a difficult time in monitoring system operation in order to detect the events which cause system failure, or performance issues, such as performance bottlenecks, for example.
Responsive to this need for diagnostic and development tools, computer engineers have developed network monitoring systems which couple into the communication channels of the network to monitor transactions between clients and servers. These systems are often hardware devices that couple into the physical layer of the network system to monitor communications. Accordingly, this requires that each physical connection between a client and a server include an interconnected hardware device. These devices monitor the data transactions that occur. By generating records of these data transactions, a system technician can attempt to identify the events which lead to the system failure and performance degradation.
Although these systems work, they require that the hardware devices are capable of detecting and recording each data transaction that occurs between the client and the server. This requires that the hardware device read each packet of data being transferred across the network to determine if the data being sent is associated with the client or the server being monitored. However, the asynchronous and rapid nature of the data transactions that occur between clients and servers renders these devices susceptible to error for failure to detect every transaction that occurs. The technician may have only a partial record of the transactions which occurred between the client and the server, and therefore, an incomplete record that is unreliable for purposes of determining the cause of the system failure and performance problems.
Other management tools exist that map a centralized system management model onto a distributed environment by implementing an agent-console architecture. In this architecture, agents continuously poll the servers and log files for the system, the network, or the applications to collect usage data and to determine if any “exception” has occurred. The console is a central management station through which the command and control functions are implemented. This architecture has several shortcomings. First, the continuous polling function employs valuable resources and degrades server performance. This is particularly true for metrics that require fine grain analysis of system activity and require constant polling. Second, the agents are at the server component level. Thus, usage, performance and exception statistics are only available at the component level and no measure is provided for end-to-end resource utilization, and no measure of the other participating components is made. Also, data gathering provisions may not be performed in real-time.
An alternative approach proposed by certain framework vendors has included an application program interface (API) to a set of resources that management tools can employ to monitor system performance. This approach requires that existing distributed applications operating on the system be edited and re-compiled to include API calls to the various system monitoring resources. Accordingly, this is a generally as highly intrusive approach to system monitoring that is dependent upon the cooperation of every vendor providing an application program running on the distributed system.
SUMMARY OF THE INVENTION
In accordance with principles of the invention is a method of monitoring a distributed computer system. Trigger events and associated data to be collected are defined. The occurrence of one of the trigger events at a client is detected while monitoring a connection between a client and a first server. Client data is collected in accordance with the one trigger event at the client. A controller is notified of the detecting of the occurrence of the one trigger event. The first server is notified of the occurrence of the trigger event. First server data is gathered by the first server, and the first server data is sent to the controller.
In accordance with anther aspect of the invention is a system for monitoring a distributed computer system. Machine executable code defines trigger events and associated data to be collected. Machine executable code detects occurrence of one of the trigger events at a client while monitoring a connection between a client and a first server. Machine executable code collects client data in accordance with the one trigger event at the client. Machine executable code notifies a controller of the detecting of the occurrence of the one trigger event. Machine executable code notifies the first server of the occurrence of the trigger event. Machine executable code gathers first server data by the first server, and machine executable code send s the first ser
Agarwal Neeraj
Briner Albert
deGroot Pieter
Doctor Murtaza
Fernandez Gary
Cardone Jason D.
Concord Communications Inc.
Hale and Dorr LLP
LandOfFree
Systems and methods for monitoring distributed applications... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Systems and methods for monitoring distributed applications..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Systems and methods for monitoring distributed applications... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3224923