Application metrics and problems

TrueSight App Visibility Manager monitors the following list of application server metrics, which can help you correlate application issues with server issues. If a metric exceeds a threshold, the tier in the Application View changes color, and the problem is displayed on the Problems tab for the tier member.

Example of the Web tier, Problems tab

Related topics

Configuring event thresholds (SLAs) for automatically discovered applications

Monitoring applications

App Visibility end-user monitoring

Event sources

Sending App Visibility health events to integrated systems

Some metrics are platform dependent (Java servers or .NET servers), but most are from both platforms.

You can configure threshold values listed as editable in the Application Configuration page.

Metrics for tiers in the Application View

MetricUnitTierDescription

Minor
threshold

Critical
threshold
EditablePlatform

Vendor
limitations

CPU usage

%

Web and Business

Java: Average percentage of CPU used by the JVM process.

NET: Average percentage of CPU used by all traced ASP.NET worker processes

7585

Contact BMC support

Java and .NETNone

Memory usage

%Web and Business

Java: Average percentage of consumed memory, consisting of heap and non-heap memory

.NET: Average percentage values of virtual and private memory that are used by all traced ASP.NET worker processes

8595Contact BMC supportJava and .NET Note: For WebSphere 7 application servers, the Memory Usage metrics sometimes display values greater than 100%. The value is displayed as it is received from WebSphere, and the same results are displayed in the JConsole tool.
Operation latency on the database%Database

Percentage of database requests that cross the latency threshold

510YesJava and .NET.NET: MS SQL 2005 and 2008, Oracle 10 and 11 
Operations with errors on the database%DatabasePercentage of database requests with errors510YesJava and .NET.NET: MS SQL 2005 and 2008, Oracle 10 and 11

Transaction latency on the server

%Web and Business

Percentage of executions for each transaction entry point that crosses the latency threshold

That is, the value represents the violation rate of each transaction entry point.

510YesJava and .NETNone
Transactions with errors on the server%Web and Business

Percentage of executions for each transaction entry point with HTTP errors

That is, the value represents the error rate of each transaction entry point.

510YesJava and .NETNone
Users experiencing operations with errors%UserPercentage of users experiencing operations with errors510YesNot applicableNone
Users experiencing slow network time%NetworkPercentage of users experiencing operations time that cross the latency threshold510YesNot applicableNone
Users experiencing slow operations%UserPercentage of users experiencing operations that cross the latency threshold510YesNot applicableNone


The following example describes how the status is determined per transaction entry point, and how this can differ for the application or server status.

Example of server status and transaction status

In a five-minute period, a server has 25 transactions without any latency violations or availability errors, and 1 transaction entry point, login.jsp, that has an exception error.

The statuses of the application and server are Critical, even though fewer than 5% of all the transactions that were executed have issues. (That is, with the default settings of 5% for Minor, 10% for Critical.)

When you drill down to the login.jsp transaction, you see the status of the transaction is Critical because it ran once and had an exception error. That is, every time the transaction ran (once) it had an error, so the percentage of transaction executions with errors for this entry point is 100%.

Metrics for tiers in the Application View, by request

The following metrics are not displayed by default; contact BMC Support to activate required metrics.

 

MetricUnitTierDescription

Platform

Vendor limitations

Deadlock

# of threadsWeb and BusinessAverage number of threads in a deadlock stateJavaNone

Memory pages per second

pages/secWeb and BusinessRate of paging memory to and from the server.NETNone

Garbage collection count

# of GC cycles 

Java: Sum of garbage collection (GC) cycles invoked by all GC algorithms

.NET: Sum of the number of generation 0 calls

Note: For WebSphere 8.5 application servers, the Garbage Collection Count metric and event displays a value that is different from the value displayed in the JConsole tool. The value in BMC Application Diagnostics is displayed as it is received from WebSphere.

Metric not displayed by default.

Java and .NETNone

Garbage collection time

ms 

Java: Sum of the time spent on GC cycles by all GC algorithms

.NET: Sum of the time spent on generation 0 calls

Java and .NETNone

Heap size - committed

MB 

Java: Sum of committed memory in all heap pools, measured after GC and as reported by JVM

.NET: Amount of virtual memory currently committed by the GC

Metric not displayed by default.

Java and .NETNone

Heap size used

MB 

Java: Sum of committed memory in all heap pools, measured after GC and as reported by JVM

.NET: Amount of virtual memory currently committed by the GC

Metric not displayed by default.

Java and .NETNone
Percent of thread pool used%Web and BusinessPercentage of the thread pool that is usedJavaWebSphere, WebLogic

Permanent generation - used

MB 

Memory used for the Permanent Generation pool

Metric not displayed by default.

JavaNone
Permanent generation - maximumMB 

Maximum memory used for the Permanent Generation pool

Metric not displayed by default.

JavaNone

Processor queue length

# of threadsWeb and BusinessNumber of ready (not running) threads in the processor queue.NETNone

Processor time

% 

Percentage of elapsed time for all processes of the OS that the processor spends to execute a nonidle thread

Metric not displayed by default.

.NETNone

Requests queued

# of requestsWeb and BusinessAverage number of ASP.NET requests waiting to be processed.NETNone

Restarted application pools

# of processes 

Sum of all restarted worker processes in all of the application pools

Metric not displayed by default.

.NETNone

Thread count

# of threadsWeb and Business

Java: Average number of threads used by the JVM process

.NET: Average number of threads used by all traced ASP.NET worker processes

Java and .NETNone
Web services duration - averagems Average time spent on web service exits, as measured on sampled requests, that are marked as exit points and as tier "w_s" in the recording profile.NETNone
Was this page helpful? Yes No Submitting... Thank you

Comments