Trace Analysis

The most common use case of the trace analysis tools is to identify user requests within a complex system landscape that have an excessive execution time. It includes analysis features across the complete solution landscape, so that a component causing a problem can be isolated and the root cause can be identified. Therefore the response times of each component involved in executing the request and the request path through the components are provided for a detailed analysis.

To analyze a trace, you check the distribution of the response time over the client, network and server. After that, you check for changes to the HTTP header (for example, with a proxy or another active network component).

We recommend that you filter out cached requests, which are usually very fast, because they are handled locally on the client browser and do not include server actions. Another common problem involves requests that transfer large amounts of data from the server to the client.

Prerequisites

Before starting a new trace, make sure that the problem to be analyzed can be reproduced.

To run a new trace, the browser requires a plug-in to enable requests leaving the client browser. In addition, the server systems must be enabled to react to these requests by switching on certain trace functionality.

Note Note

The current version of the plug-in is available for download as an attachment to SAP Note 1435190.

Alternatively, you can download the plug-in directly from the trace analysis application. You can find the links in the Options section of the analysis application.

End of the note.

Before starting the trace, you must enable tracing on the server; the related function is located in the area Trace Enabling of the trace analysis tool. This area displays the systems contained in the selected solution. To enable a system to run traces, select the corresponding row in the table and choose Enable All. In the column State, a green icon indicates that a system has been enabled to run traces. In the Options area, you can define for how long the trace enabling is valid. By default, tracing is automatically disabled after 30 minutes.

Procedure

The trace analysis is divided into two parts:

A trace is run and recorded.
The recorded trace is analyzed.

Run and Record Trace

To run a trace, activate the client plug-in before performing the action to be analyzed. You can activate the plug-in by runnning the executable file SAPIEPlugin.exe. An empty browser window opens with the deactivated plug-in user interface.

Note Note

By entering the browser URL, you can navigate to where the error occurred without having to write trace data.

End of the note.

Enter a name for the business transaction and a name for the transaction step. In both cases you can choose any name, the names are only used to identify the trace in the analysis tool later.

In the next field, select the trace level. The following table shows how different trace levels are used for different analytical purposes:

Trace Level	Purpose
none	No server traces are written. This is helpful if you want to trace only the end-user client experience.
low	Only the HTTP logs are written. This is useful for HTTP analysis only because it provides front-end response times, server response times (HTTP logs) and the resulting network times.
medium	Provides the most important information and is used for a performance trace.
high	Used for tracing functional issues, technical correctness, and data consistency. Depending on the location, this could have a significant impact on the response time, and should not be used in a productive environment.

To enable the automatic upload of data to the client, enter the host name of the SAP Solution Manager installation and the HTTP port in the user interface. If there is no HTTP connection between the client and the SAP Solution Manager host, the trace data can be uploaded into the analysis tool manually at a later time.

After making all the settings, activate the plug-in by choosing Start. As of this point, all requests will be instrumented and measured by the plug-in. During the runtime of the transaction, the plug-in displays:

Bytes sent and received
Number of requests and responses
Maximum number of physical connections
Currently active connections.

When the transaction has finished, the overall response is displayed in the browser.

It is now possible to trace additional transaction steps. To do this, choose New Step and enter a descriptive name for the next transaction step.

To stop the trace, choose Stop Transaction. The trace data is then automatically transferred to the analysis tool in the SAP Solution Manager. If the automatic transfer is not successful, an error message appears and you can upload the data manually. The relevant file is the BusinessTransaction.xml, which is located in the folder logs and subfolder <timestamp>_<transactionname> of the plug-in directory.

To leave the browser window and the plug-in, choose Exit.

The next step is to analyze the trace.

Analyze a Trace

To start a trace analysis, go to the area Root Cause Analysis:

In the top level navigation, choose End to End Analysis.
In the Detailed Selection area, select the system(s) that you want to analyze.
In the header area of the Detailed Selection list, click Trace Analysis.
The Detailed Selection screen now displays the Trace Analysis views with one tab for each analyzed system. Note the following on the Trace Analysis panel(s):
- If the client trace data was automatically transferred, the descriptive name of the transaction is displayed as a new entry in the section Trace Analysis at the top of the table. You might need to choose Refresh.
- To manually upload the file, click the entry field Upload BusinessTransaction.xml in the lower part of the screen.
Select the correct server trace data to be uploaded to the analysis tool.
Make sure that the correct solution is selected.
Select the new trace in the table.
Verify the solution and confirm the dialog box.

To display the transaction steps of a business transaction, select a Business Transaction. The screen area Transaction Steps is displayed.

To display the analysis view, select a transaction step and click Display. The view contains summarized information on the trace and the most important metrics collected during the trace. The screen is divided into four major parts:

Client Trace Summary: Contains all the values measured by the SAP HTTP plug-in.
Time (accumulated): The network time per HTTP request between the browser and the first HTTP server component is calculated as the difference between the browser response time and the server time. The distribution of the time intervals is displayed in a pie graph.
HTTP Status Codes: Contains all the returned status codes for all requests. The distribution of the status codes is displayed in a pie chart.
Transaction Step Details: Lists all instances and server nodes that are involved with direct HTTP communication.

The accumulated times allow you to see at a glance whether the performance problem is on the client, in the network, or in the server.

Note Note

The network time displayed in the pie chart is calculated by means of the difference between client response time and the server response time, and not measured on the network layer. Therefore, the network time must be seen as an approximation.

End of the note.

Note Note

The client response time is measured inside the browser. It starts when the first byte of an HTTP request leaves the browser and ends when the last byte of an HTTP response arrives. As a result, the client response time includes time that was spent by the browser itself as well as the time spent by other processes running on the operating system (for example, virus scan, firewall).

End of the note.

Result

You can display data from which you can perform further analyses.