Root cause analysis of server system performance degradations

Details Root cause analysis of server system performance degradations Root cause analysis of server system performance degradations

2001-10-19

2004-05-18

Beausoliel, Robert (Department: 2184)

Error detection/correction and fault detection/recovery

Data processing system error or fault handling

Reliability and availability

C714S025000, C714S026000, C702S186000

Reexamination Certificate

active

06738933

ABSTRACT:

FIELD OF THE INVENTION
The present invention relates to software tools and services for testing, monitoring and analyzing the operation of web-based and other transactional servers.
BACKGROUND OF THE INVENTION
A variety of commercially-available software tools exist for assisting companies in testing the performance and functionality of their web-based transactional servers and associated applications prior to deployment. Examples of such tools include the LoadRunner®, WinRunner® and Astra QuickTest® products of Mercury Interactive Corporation, the assignee of the present application.
Using these products, a user can record or otherwise create a test script which specifies a sequence of user interactions with the transactional server. The user may also optionally specify certain expected responses from the transactional server, which may be added to the test script as verification points. For example, the user may record a session with a web-based travel reservation system during which the user searches for a particular flight, and may then define one or more verification points to check for an expected flight number, departure time or ticket price.
Test scripts generated through this process are “played” or “executed” to simulate the actions of users—typically prior to deployment of the component being tested. During this process, the testing tool monitors the performance of the transactional server, including determining the pass/fail status of any verification points. Multiple test scripts may be replayed concurrently to simulate the load of a large number of users. Using an automation interface of the LoadRunner product, it is possible to dispatch test scripts to remote computers for execution.
The results of the test are typically communicated to the user through a series of reports that are accessible through the user interface of the testing tool. The reports may contain, for example, graphs or charts of the observed response times for various types of transactions. Performance problems discovered through the testing process may be corrected by programmers or system administrators.
A variety of tools and services also exist that allow web site operators to monitor the post-deployment performance of their web sites. For example, hosted monitoring services now exist which use automated agents to access a web site at regular intervals throughout the day. The agents measure the time required to perform various web site functions, and report the results to a server provided by Keynote Systems. The owner or operator of the web site can access this server using a web browser to view the collected performance data on a city-by-city or other basis. Other types of existing monitoring tools include log analysis tools that process access logs generated by web servers, and packet sniffing tools that monitor traffic to and from the web server. Further, using the LoadRunner ActiveTest service of Mercury Interactive Corporation, companies can load test their web sites and other systems over the Internet prior to deployment.
SUMMARY
A significant problem with existing monitoring tools and services is that they often fail to detect problems that are dependent upon the attributes of typical end users, such as the user's location, PC configuration, ISP (Internet Service Provider), or Internet router. For example, with some web site monitoring services, the web site operator can monitor the web site only from the agent computers and locations made available by the service provider; as a result, the service may not detect a performance problem seen by the most frequent users of the system (e.g., members of a customer service department who access the web site through a particular ISP, or who use a particular PC configuration).
Even when such attribute-specific problems are detected, existing tools and services often fail to identify the specific attributes that give rise to the problem. For example, a monitoring service may indicate that web site users in a particular city are experiencing long delays, but may fail to reveal that the problem is experienced only by users that access the site through a particular router. Without such additional information, system administrators may not be able to isolate and correct such problems.
Another significant problem with existing tools and services is that they do not provide an adequate mechanism for monitoring the current status of the transactional server, and for promptly notifying system administrators when a problem occurs. For example, existing tools and services typically do not report a problem until many minutes or hours after the problem has occurred. As a result, many end users may experience the problem before a system administrator becomes aware of the problem.
Another significant problem with prior tools and services is that they generally do not provide a mechanism for identifying the source of performance problem. For instance, a web site monitoring service may determine that users are currently experiencing unusually long response times, but typically will not be capable of determining the source of the problem. Thus, a system administrator may be required to review significant quantities of measurement data, and/or conduct additional testing, to pinpoint the source or cause of the detected problem.
The present invention addresses these and other problems by providing a software system and method for monitoring the post-deployment operation of a web site system or other transactional server. In a preferred embodiment, the system includes an agent component (“agent”) that simulates the actions of actual users of the transactional server while monitoring and reporting the server's performance. In accordance with one aspect of the invention, the agent is adapted to be installed on selected computers (“agent computers”) to be used for monitoring, including computers of actual end users. For example, the agent could be installed on selected end-user computers within the various offices or organizations from which the transactional server is commonly accessed. Once the agent component has been installed, the agent computers can be remotely programmed (typically by the operator of the transactional server) using a controller component (“controller”). The ability to flexibly select the computers to be used for monitoring purposes, and to use actual end-user computers for monitoring, greatly facilitates the task of detecting problems associated with the attributes of typical end users.
In accordance with another aspect of the invention, the controller provides a user interface and various functions for a user to remotely select the agent computer(s) to include in a monitoring session, assign attributes to such computers (such as the location, organization, ISP and/or configuration of each computer), and assign transactions and execution schedules to such computers. The execution schedules may be periodic or repetitive schedules, (e.g., every hour, Monday through Friday), so that the transactional server is monitored on a continuous or near-continuous basis. The controller preferably represents the monitoring session on the display screen as an expandable tree in which the transactions and execution schedules are represented as children of the corresponding computers. Once a monitoring session has been defined, the controller dispatches the transactions and execution schedules to the respective agent computers over the Internet or other network. The controller also preferably includes functions for the user to record and edit transactions, and to define alert conditions for generating real-time alert notifications. The controller may optionally be implemented as a hosted application on an Internet or intranet site, in which case users may be able to remotely set up monitoring sessions using an ordinary web browser.
During the monitoring session, each agent computer executes its assigned transactions according to its assigned execution schedule, and generates performance data that indicates one or more characteristics of the transac

Affiliated with

Also associated with

Rating

Root cause analysis of server system performance degradations does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Root cause analysis of server system performance degradations, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Root cause analysis of server system performance degradations will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3238569