Cisco Technology, Inc.

Appdynamics, Inc.

Vinay Srinivasaiah

Manoj Acharya

Bhaskar Sunkara

Binil David Thomas

Jyoti Bansal

Awais Ahmed

Sanjay Nagaraj

Suraj Puvvada

Bradley Winslow

Jad Naous

Adam Ross Cath

Puneet Anand

Ashwin Jayaprakash

Jonathan Newcomb Swirsky Whitney

William Drake Pijewski

Ryan Nicholas TerBush

Ed Rowe

Bhaskar Sankara

Juhwan Jeong

Peter Howard Holditch

In one embodiment, a method for automatic application data collection is disclosed. The method illustratively comprises determining a stack trace associated with one or more business transactions, wherein at least one application is configured to perform at least part of the one or more business transactions; identifying, based on the stack trace, one or more methods of the at least one application that were invoked during performance of the one or more business transactions; selecting, from the identified one or more methods, a set of the one or more methods by applying at least one scoring heuristic to the identified one or methods; and configuring one or more monitoring agents executing on the at least one application to instrument the selected set of the one or methods at one or more instrumentation points, wherein application performance data is generated by instrumentation of the one or more instrumentation points.

Automatic application data collection for potentially insightful business values

A system monitors a network or web application provided by one or more distributed applications and provides data for each and every method instance in an efficient low-cost manner. Agents may monitor the performance of the distributed application by the web services and report monitoring data as runtime data to the remote server, for example a controller. The controller may analyze the data to identify one or more performance issues or 'hot spot' methods based on current or past performance, functionality, content, or business relevancy. Instructions and/or configuration information may be transmitted by the controller to the agents that correspond to a particular business transaction portion associated with a hot spot. The portions are then monitored to collect data associated with the hot spot and the hot spot data is reported back to the controller.

Automatic capture of detailed analysis information based on remote server analysis

In one aspect, a system for providing automated root cause analysis of a monitored business transaction is disclosed. The system includes a processor; a memory; and one or more modules stored in the memory and executable by a processor to perform various operations. For example, a performance issue associated with the business transaction running in a monitored environment can be detected and automated root cause analysis can be provided to present to a user a chain of entities associated with detected performance issue. A score can be provided for each node of entity in the chain to indicate which node of entity is likelihood to be a root cause candidate for the performance issue.

Automated model based root cause analysis

Asynchronous handoffs between threads and other software components may be automatically detected, and the corresponding working objects may be tracked. The system may report monitoring information for an overall transaction that includes the original request and corresponding asynchronous requests. Automatically detecting asynchronous requests may include instrumenting a virtual machine, such as a Java Virtual Machine (JVM), to detect the creation of thread handoff objects and the object and/or thread execution. Thread handoff objects may automatically tracked, tracked based on data learned over time, tracked based on user input, and otherwise configured. In some embodiments, after detecting the creation of a thread handoff object, an identification of the object of the call may be identified as being tracked in another server or application.

Automatic asynchronous handoff identification

Asynchronous operations associated with a request such as threads, runnable elements, callable elements, and other invokable objects are tracked to determine the metrics about the request and operations. The present technology tracks the start and end of each asynchronous operation and maintains a counter which tracks the currently pending or executing asynchronous operations. By monitoring the request, the start and end of each asynchronous operation associated with the request, and the number of asynchronous operations currently executing, the present technology may identify the end of a request by identifying when the last asynchronous operation associated with the request ends. In some instances, the present technology identifies the end of a request when a counter which tracks the number of asynchronous operations executing reaches a value of zero after the first asynchronous operation has already begun. Tracking these operations allows the present technology to aggregate and report useful performance data about these requests.

Asynchronous processing time metrics

The system identifies objects that cause thrashing behavior in garbage collection. A garbage collection process may be monitored for a period of time. Over that period of time, a number of objects may be observed to be collected by the garbage collection process. Data may be collected for those objects and a subset of those objects may be determined to be suspicious based on data collected for each object. The suspicious objects may then be reported as causing garbage collection thrashing.

Automatic garbage collection thrashing monitoring

The present technology correlates tasks between multiple threads of an application. A payload consisting of a task object may be identified as part of a method. Identification of the task object may be performed using instrumentation or some other method. Once the task object is identified, the payload may be instrumented to track the task object or payload and correlate the task object with other executing threads. Instrumenting the payload may include determining the constructor of the task object and instrumenting the task object constructors to determine what creates the task object. Once the source of a task object or payload is determined, the current thread having the task object as a parameter and another thread which creates the task may be tied together as part of a business transaction.

Automatic task tracking

The present technology utilizes agents to monitor and report data from Java virtual machines (JVM) to a controller as part of application performance monitoring. When a JVM is loaded, code defining an interface for agents is loaded as well. A determination may be made as to whether the loaded agent implements the interface defined at the JVM. If the loaded agent does not implement the interface, for example if it is missing one or more methods defined by the interface, the agent class may be modified to define the missing methods. The modification to the agent class may be made after compilation but before the class is loaded into the JVM.

Dynamic agent delivery

A system monitors a client request and a backend business transaction performed in response to the request and correlates the data generated from monitoring the request and business transaction. A request may be monitored in detail to determine several metrics. The metrics may include server connection time, response receipt completion time, document download time, document processing time, and page rendering time. The metrics may be determined for an entire content page as well as individual page elements, such as Iframes and elements constructed from AJAX techniques.

Business transaction correlation with client request monitoring data

A system monitors a network or web application provided by one or more distributed applications and provides data for each and every method instance in an efficient low-cost manner. The web application may be provided by one or more web services each implemented as a virtual machine or one or more applications implemented on a virtual machine. Agents may be installed on one or more servers at an application level, virtual machine level, or other level. The agent may identify one or more hot spot methods based on current or past performance, functionality, content, or business relevancy. Based on learning techniques, efficient monitoring, and resource management, the present system may capture data for and provide analysis information for outliers of a web application with very low overhead.

Growing community of inventors

San Carlos, CA, United States of America

Vinay Srinivasaiah