The present invention relates to JDBC subsystems, especially diagnostic, monitoring and debugging components for JDBC subsystems.
Java Database Connectivity (JDBC) is an API that provides connectivity to a wide range of databases as well as access to other types of tabular data sources, such as spreadsheets or flat files. With a JDBC technology enabled driver, users can connect to all types of corporate data in a heterogeneous environment. To use the JDBC API with particular database management system, a JDBC technology based driver is used to mediate between JDBC technology and the database.
The JDBC subsystem can be associated with an application server such as the WebLogic™ Server available from BEA Systems, Inc. of San Jose, Calif.
Multiple connection pools and multipools can be used. In one embodiment, the JDBC subsystem 206 parses a database requests to produce statements to be sent to the database management system (DBMS). In order to speed up these requests, a previously parsed statement can be stored in a prepared statement cache 210. The prepared statement cache 210 can include prepared statements, such as prepared statements 210a, 210b, 210c and 210d. The JDBC subsystem 206 can check the prepared statement cache to see whether a request is the same as a previously received request. This is quite likely when the client, such as application 202, sends the same request multiple times.
In one embodiment of the present invention, the JDBC subsystem contains, or is associated with, a profiling component 212. In one embodiment, the profiling component 212 is configured to maintain profile information 214 concerning components of JDBC subsystem 206. The profile information 214 can be derived from the statistics 220.
In one embodiment, the profile component 212 maintains profile information 214 concerning the connection pool 208. The profile information 214 can include a connection usage profile indicating who currently has a pool connection. Such information is useful for administrators of the application system. In one embodiment, administrators can access the JDBC subsystem 206 using an administration console 216. In one embodiment, the connection usage profile indicates when a connection was reserved.
In one embodiment, the profiling component contains profile information including a wait profile indicating who is waiting for a connection. The wait profile can indicate how long the person has been waiting for this connection. The wait profile is useful for checking the operation of the system.
The profiling information 214 can also include a wait failure profile indicating who waited for a connection. The wait failure program can also indicate how long a client waits for a connection. Such information is useful when analyzing the operation of the JDBC subsystem 206.
The profiling information 214 can include connection leak profile information indicating who reserved the connection but did not use it. In one embodiment, applications or internal clients that hold onto a connection longer than a predetermined time, such as 30 seconds, can be considered to have held onto the connection in error. Maintaining the connection leak profile allows the administrator to see which applications or internal clients are failing in this manner.
The profiling component can maintain profile information concerning the prepared statement cache 210. The profile information can include a cached statement profile indicating who has added prepared statements to the prepared statement cache 210. The profile information can include a statements executed profile indicating who executed cache statements and how long the execution of the cached statements took. Such information is useful for an administrator to adjust the prepared statement cache 210. For example, a prepared statement cache 210 may be adjustable in terms of the number of prepared statements stored. Analysis of the profile information allows the administrator to select the correct size of the prepared statement cache 210.
The JDBC subsystem 206 can include or be associated with a statistics monitoring component 218. The statistics monitoring component 218 can be configured to maintain statistics 220. Statistics 220 can included statistics concerning the connection pool 208 and prepared a statement cache 210. In one embodiment, the statistics 220 indicate the cumulative number of requests to reserve a connection from the connection pool 208. In one embodiment, the statistics include the cumulative number of times a reserve request fails. The statistics can include an average time a connection is in use by a client. In one embodiment, the statistics include the average time the client waited for a connection. In one embodiment, the statistics include the percentage of time connections in the connection pool were used. In one embodiment, the statistics include the percentage of time all of the connections in the pool were in use. In one embodiment, the statistics include the cumulative number of requests to reserve a connection of a pool that had to wait. In one embodiment, the statistics include a cumulative number of times a requests to reserve that waited that failed to get a connection.
The system statistics can concern the prepared statement cache 210. The statistics can include the cumulative number of times that the cache 210 is accessed. The statistics can include the cumulative number of statements are added to the cache. The statistics can include the cumulative number of statement discarded from the cache. The system can also include the cumulative number of statements in the cache. The statistics can be used by an administrator to monitor the operation of the JDBC subsystem 206.
In one embodiment, the JDBC subsystem 206 contains, or is associated with, a debugging component 220. The debugging component 222 can debug accesses to the connection pool 208 and/or the prepared statement cache 210. The debugging component 222 can store trace information 224 concerning the debugging. The debugging can be selectable on and off.
In one embodiment, the debugging component is configured to trace methods involved in a JDBC driver. The debugging component can be selectable to enable or disable the traces. The debugging component can trace the arguments of methods invoked in the JDBC driver. The debugging component can trace the return values of the methods invoked in the JDBC driver, errors of message invoked in the JDBC driver and exceptions of the messages invoked in the JDBC driver. The debugging component can be configured to trace operations invoked in the connection pool.
The debugging component can trace connection reserve operations, connection release operations, application requests to get connections, and application requests to close connections.
The debugging component can be configured to trace operations involved in a prepared statement cache. The debugging component can trace cache hits, cache misses, cache statement additions, and cache statement releases.
In one embodiment, the debugging statistics and profiling data can be switched on and off using configuration information in the diagnostics configuration memory 230. The diagnostics configuration memory 230 can be a MBean which stores the diagnostic configuration information for JDBC subsystem.
One embodiment of the present invention, is a method for operating a JDBC subsystem 301 comprising accessing a JDBC driver 308 at the request of an application 302. During the accessing step, it is checked whether the application has a diagnostic callback. If the application has a diagnostic callback, the diagnostic callback is started.
The diagnostic callback can then execute and then do operations such as obtaining profile statistic information 320.
In one embodiment, the callbacks can be provided by the application to do the diagnostic functions desired by the callback. In one embodiment, the diagnostic callback is called upon an exception. The diagnostic callback can be called before executing a method of the JDBC driver, after executing a method in the JDBC driver and upon an exception in the method.
The diagnostic callback can profile the JDBC driver. The callback can monitor methods of the JDBC driver being executed. The callback can monitor exceptions thrown by the JDBC driver.
The diagnostic callback can be interface that it can be implemented by the application 302. The diagnostic callback can be registered with the JDBC subsystem using the JDBC system callback registration 304. The registration can indicate whether a callback is to be run as well as indicating the point to the code that the callback code. The JDBC subsystem can check whether the diagnostic callback is registered during the accessing step.
One embodiment of the present invention comprises adding JDBC configuration information into a diagnostic image. Upon an event, compiling diagnostic information from the JDBC subsystem can be compiled allng with other information about the state of the softwareabout the state of server software into a data stream; and compressing the data stream.
In one embodiment, a diagnostic image is not created for each failure. The creation of multiple diagnostic images can be expensive because of large size of the diagnostic images. A timer can be used to inhibit the collection of multiple diagnostic images.
One embodiment of the present invention includes software including code to, upon an event, compile diagnostic information about the state of server software into a data stream; and code to compress the data stream.
One embodiment of a diagnostic image system is described in the application entitled “Diagnostic Image”, application Ser. No. 11/133,566 filed May 20, 2005 incorporated herein by reference [Attorney Docket No. BEA-01695US0].
One non-limiting example of a JDBC Diagnostics System is described below:
STATISTICS—User can be able to view snapshots of values of attributes of configured JDBC objects that are either current or recently recorded.
Following new statistics can be made available for connection pools:
Cumulative number of requests to reserve a connection from pool
Cumulative number of times reserve requests failed
Average time a connection is in use by client
Average time a client waited for a connection
Percentage of connections in pool currently in use
Percentage of time all connections in pool were in use
High watermark of current capacity of pool
Cumulative number of requests to reserve a connection from pool that had to wait
Cumulative number of times reserve requests that waited, failed to get a connection
The Following new statistics can be made available for the prepared statement cache:
Cumulative number of times cache accessed
Cumulative number of statements added to the cache
Cumulative number of statements discarded from the cache
Current number of statements in cache
PROFILES—User can be able to view data aggregated and/or derived from values of attributes of configured JDBC objects. This can help the user in understanding the usage patterns of the JDBC objects by the application.
The following information about connection pools can be profiled:
Connection Usage (who currently has a pool connection, when did they reserve it)
Wait (who waited for a connection, how long)
Wait Failure (who waited for a connection, how long)
Connection Leaks (who had reserved it)
The following information about the prepared statement cache can be profiled:
Cached statements (who added them)
Statements executed (who executed them, how long did execution take)
EXEMPLARY PROFILE INFORMATION INTERFACE—The collected profile information can be stored in an archive, such as the WLDF Archive in records that are instances of the type weblogic.jdbc.extensions.ProfileDataRecord.
APPLICATION CALLBACKS—The Application can register callbacks that can get invoked before and after invocations of methods in the JDBC driver.
In one embodiment, the application can provide an implementation of the interface “weblogic.jdbc.extensions.DriverInterceptor” and register it with the JDBC subsystem using the element “driver-interceptor” of the descriptor.
JDBC subsystem can invoke methods “preInvokeCallback( )”, “postInvokeExceptionCallback( )” and “postInvokeCallback( )” of the registered callback before and after invoking any method inside the JDBC driver.
Applications can implement this interface to profile the usage of the JDBC driver and monitor things like:
Methods being executed
Any exceptions thrown
Time spent inside the driver executing methods
Exemplary Application Callbacks Interface
DEBUGGING—Application can be able to analyze the usage of JDBC objects such as connection pools, prepared statement caches or JDBC drivers. In one embodiment, this can be done by enabling attributes of the ServerDebugMBean. In one embodiment, this can be done by application server developers personnel, rather than the customer themselves.
In one embodiment, the efficacy of existing JDBC-related attributes “JDBCSQL” and “JDBCConn” in ServerDebugMBean are improved:
JDBCSQL—print out trace all methods invokved in the JDBC driver, their arguments and return values, errors and exceptions.
JDBCConn—trace all connection reserve and release operations in connection pools as well as all application requests to get or close connections
Following new attributes can be added to ServerDebugMBean:
JDBCStatementCache—trace all cache hits and misses, additions and deletions
JDBCDriverLogging—enable JDBC driver level logging. This will deprecate the existing mechanism of enabling JDBC driver level logging using attributes in ServerMBean. —JDBCRMI—Debugging at the RMI level—intended primarily for CCE/Engineering use.
JDBCInternal—Internal debugging information—intended for CCE/Engineering use.
Functionality provided by debugging systems can be used to output all information generated by enabling the abovementioned attributes.
As facilitated by the debugging system, a new Debug Scope called “weblogic.jdbc” can be created and all abovementioned new and existing ServerDebugMBean attributes cab be tagged as being part of this Debug Scope. Existing ServerDebugMBean attribute “JTAJDBC” cab also be tagged as being part of weblogic.jdbc scope.
DIAGNOSTIC IMAGE—In one embodiment, the application can be able to obtain a Diagnostic Image of the JDBC subsystem.
REQUEST DYEING—In one embodiment, the user can trace the flow of an individual (typically “dyed”) application request through the JDBC subsystem.
The foregoing description of preferred embodiments of the present invention has been provided for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many embodiments were chosen and described in order to best explain the principles of the invention and its practical application, thereby enabling others skilled in the art to understand the invention for various embodiments and with various modifications that are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the claims and their equivalents.
This application claims priority to U.S. Provisional Application No. 60/725,353 entitled “JDBC Monitoring and Diagnostics Enhancements” filed Oct. 11, 2005 [Attorney Docket No. BEAS-01767US0].
Number | Date | Country | |
---|---|---|---|
60725353 | Oct 2005 | US |