Freigeben über


Procedures: Using the Data Viewing Features

The procedures in this section demonstrate the use of numerous Message Analyzer features that are described in the Viewing Message Data section. They are intended to serve as a cross section of the many ways in which you can use Message Analyzer viewer features and other integrated functions. As viewing message data and analyzing it are closely related, these procedures demonstrate the use of various viewers and tools that manipulate trace results data for analysis purposes. For additional information on the analysis tools that Message Analyzer provides, see Analyzing Message Data.

Important

Although these procedures demonstrate the use of Message Analyzer capabilities in some basic analysis scenarios, they are only a sampling of what you can accomplish with Message Analyzer, given that you can also apply the methodologies described here to many other scenarios.


Procedure Overviews
A brief description of each procedure is included here for review, as follows.


Apply Gradient Style Color Rules — provides an example of how to utilize gradient style Color Rules to quickly flag messages that meet the filtering criteria of multiple Color Rules.

Apply a Built-In View Layout — provides an example of a built-in view Layout that presents a data column configuration that is useful for diagnosing TCP messages when applied, while also automatically grouping messages by IP conversations and ports (Group operations on the Network and Transport columns, respectively) to enhance diagnostic capabilities and perspectives.

Perform Data Grouping Operations — provides several examples of data grouping operations that demonstrate how you can filter and consolidate data from designated Analysis Grid viewer columns and reorganize them into separate groups of common properties that greatly enhance your ability to analyze data and resolve issues.

Perform Top-Level Summary Analysis — provides an example of how to use the Protocol Dashboard viewer to obtain top-level summaries at a glance for a set of trace results.

Perform Interactive Analysis with Data Viewers — illustrates a simple method for using the Protocol Dashboard and Analysis Grid viewers together in an interactive manner to enhance data analysis perspectives.

Apply Viewpoints to Trace Data — provides an example that shows you how to use the Message Analyzer Viewpoints feature, which enables you to examine data from the viewpoint of a protocol, where the messages of a specific viewpoint protocol are displayed at top-level in the Analysis Grid viewer with no message layers above them.

Apply a Time Filter to Trace Results — provides an example that shows you how to apply a Time Filter to a set of trace results, so that you can view data in a selected window of time.

Drive Analysis Grid Viewer and Tool Window Interactions — provides an example that demonstrates interaction between the Analysis Grid viewer and various tool windows, such as the Message Data, Field Data, Message Stack, Details, and Diagnostics Tool Windows.

Create an Alias for a Data Field Value — provides an example that demonstrates how to simplify data analysis by creating an Alias that substitutes for a cryptic field value.

Create a Union of Two Data Fields — provides an example that demonstrates how to create a Union that correlates/combines two data fields with similar values but different names into a single new field that is specified by the Union configuration.

Procedures: Using the Data Filtering Features — see this procedural topic for extensive coverage of different ways to apply a view Filter.

Important

If you have not logged off Windows after the first installation of Message Analyzer, please log off and then log back on before performing these procedures. This action ensures that in all subsequent logons following installation, your security token will be updated with the required security credentials from the Message Capture Users Group (MCUG). Otherwise, you will be unable to capture network traffic in Trace Scenarios that use the Microsoft-PEF-NDIS-PacketCapture provider, Microsoft-Windows-NDIS-PacketCapture provider, or the Microsoft-PEF-WFP-MessageProvider, unless you start Message Analyzer with the right-click Run as administrator option.

Apply Gradient Style Color Rules

In the procedure that follows, you will apply the built-in IPv4 Right Gradient and TCP left gradient Color Rules to a Link Layer trace that captured data with the Local Network Interfaces Trace Scenario that uses the Microsoft-PEF-NDIS-PacketCapture provider (available on Windows 7, Windows 8, and Windows Server 2012 operating systems).

Note

If your machine is running the Windows 8.1 or later operating system, you can capture data in this example with the Local Network Interfaces Trace Scenario that uses the Microsoft-Windows NDIS-PacketCapture provider.

This procedure demonstrates a simple way to expose TCP messages that have an IPv4 Network Layer in the message stack. The Color Rule that is used in the procedure also serves as an example of how you might design multiple gradient-style Color Rules with visually-coordinated opposite facing gradients, which you can then use as a troubleshooting mechanism to quickly identify message stack components at a glance.


More Information
To learn more about the concepts upon which this example procedure is based, see Using and Managing Color Rules.


To identify Transport and Network Layer messages with gradient-style Color Rules

  1. From the Start menu, Start page, or task bar of your computer, click the Microsoft Message Analyzer icon to launch Message Analyzer.

    To ensure that you have access to all features, run Message Analyzer as an Administrator.

  2. Click New Session on the Start Page to open the New Session dialog.

  3. Under Add Data Source in the New Session dialog, click the Live Trace button to display the Live Trace tab along with the associated session configuration features that it contains in the New Session dialog.

  4. In the Network category of the Select Scenario drop-down list on the Live Trace tab, click the Local Network Interfaces Trace Scenario.

    If your operating system is Windows 7, Windows 8, or Windows Server 2012, the ETW Providers list on the Live Trace tab is populated with the Microsoft-PEF-NDIS-PacketCapture provider Name and Id (GUID). Otherwise, for the Windows 8.1, Windows Server 2012 R2, Windows 10 operating system, or later, the Microsoft-Windows-NDIS-PacketCapture provider information displays.

  5. Click the Start button in the New Session dialog to automatically select the default data viewer and start capturing data. Assuming that you have not changed the default data viewer in the Default Profile pane on the Profiles tab of the global Options dialog, the default viewer will be the Analysis Grid.

    Message Analyzer should immediately begin capturing data and accumulating it in the Analysis Grid viewer.

  6. While Message Analyzer is capturing data, attempt to reproduce any conditions that are related to a particular TCP or IPv4 issue you are trying to isolate, for example, network connection or packet loss problems.

  7. Stop the trace at a suitable point by clicking the Stop button on the global Message Analyzer toolbar.

Tip

You can temporarily suspend tracing operations by clicking the Pause/Resume button and you can resume tracing by clicking the Pause/Resume button again.

  1. Click the Color Rules drop-down list on the Analysis Grid viewer toolbar, and then under the Network category of the drop-down that displays, select the TCP left gradient and IPv4 Gradient Right Color Rules.

    All top-level TCP messages or other top-level messages that have TCP in the origins tree are highlighted with the light blue left-to-right gradient Color Rule style. Also, all TCP messages that have an IPv4 network layer are highlighted in the olive green right-to-left gradient Color Rule style, thus enabling you to easily view messages that meet the filtering criteria of both the applied Color Rules.

Note

To isolate either TCP or IPv4 messages at top-level to further enhance analysis, you can apply a TCP or IPv4 viewpoint as appropriate from the Filtering toolbar.

Apply a Built-In View Layout

In the procedure that follows, you will apply the built-in TCP Deep Packet Analysis with Absolute Sequence Number Grouping view Layout to trace data that is displayed in the Analysis Grid viewer. This view Layout has a column layout configuration that contains various TCP fields, the values of which can be important when diagnosing TCP issues. The columns that hold TCP field data include DestinationPort, SourcePort, PayloadLength, SequenceNumber, AcknowledgementNumber, and WindowScaled columns. In addition, a TimeDelta field is also included to display the running time for captured messages. The predefined Layout also includes groupings of Network and Transport columns that present the details of the IP conversations that took place on corresponding TCP ports within a trace. Note that the Network and Transport columns were removed after the Grouping operation, but before this Layout was saved in the default Layout Library item collection.


More Information
To learn more about the concepts upon which this example procedure is based, see Applying and Managing Analysis Grid Viewer Layouts.


To apply a built-in View Layout for TCP diagnosis

  1. Perform steps 1 through 7 of the procedure To identify Transport and Network Layer messages with gradient-style Color Rules to start and stop a Message Analyzer Live Trace Session that uses the Local Network Interfaces Trace Scenario.

  2. Click the Layout drop-down list on the Analysis Grid toolbar and then click the TCP Deep Packet Analysis with Absolute Sequence Number Grouping item in the drop-down list that appears.

    The new column configuration displays and the data is grouped into Network, Transport, and Sourceport groups, as indicated by corresponding labels above the tree grid. The data groups are also organized such that the Transport nodes are nested within the Network nodes, and the Sourceport nodes are nested within those. Note that the Network conversations can use either IP or Ethernet addresses.

  3. Expand a particular Network node to expose the Transport node it contains.

    The exposed Transport node provides an indication of the number of messages that it contains, along with the source and destination TCP ports over which IP or Ethernet conversations took place.

  4. Expand the Transport node to display the TCP messages, so that you can examine the TCP field data. If you are dealing with loss of packets, you might check the WindowScale field for low values.

    Note

    For convenience of viewing the TCP column data, you can alter the data columns that will horizontally scroll by right-clicking the column that you want as the first scrollable column and then select the Freeze Columns to Left command in the context menu that appears.

  5. Expand the Sourceport nodes, so that you can view the messages that transited each TCP port.

  6. Repeat steps 3, 4, 5 for other Network, Transport, and Sourceport nodes as appropriate.

  7. To obtain a different perspective on the data, you can drag the Transport group label and drop it to the left of the Network group label.

    The data is now grouped and organized with Network nodes nested within the Transport nodes.

    Note: You can drag and drop any of the Group labels into any position that you want, to change the way messages are hierarchically organized.

Note

To restore the default Layout, click the Layout drop-down list on the Analysis Grid viewer toolbar, click the Manage Layouts item, and then click the Restore Application Default Layout command that displays in the submenu that appears.

Perform Data Grouping Operations

In the procedure that follows, you will execute the Group command on various Analysis Grid viewer data columns, including the ContentType, Transport, Source or Destination, and Diagnosis data columns. The grouping operations will enable you to quickly determine the object types being requested by your web browser, assess the heaviest port traffic, determine the IP addresses carrying the most traffic, and examine grouped diagnosis messages types, respectively. In this procedure, the Analysis Grid viewer will be populated with message data that you capture with the Microsoft-PEF-WFP-MessageProvider in the Loopback and Unencrypted IPSEC Trace Scenario and the focus will be on Application Layer (HTTP) and Transport Layer messages.


More Information
To learn more about the concepts upon which this example procedure is based, see Using the Analysis Grid Group Feature.


To perform multiple data grouping operations for analysis

  1. Perform steps 1 through 3 of the procedure To identify Transport and Network Layer messages with gradient-style Color Rules, to start Message Analyzer and open the New Session dialog for Live Trace Session configuration.

  2. In the Network category of the Select Scenario drop-down list on the Live Trace tab of the New Session dialog, click the Loopback and Unencrypted IPSEC Trace Scenario. Alternatively, click the Loopback and Unencrypted IPSEC Trace Scenario in the Favorite Scenarios list that is accessible from the Message Analyzer File menu.

    The Microsoft-PEF-WFP-MessageProvider information displays in the ETW Providers list on the Live Trace tab, which includes the provider Name, GUID, and a Configure link that opens the Advanced Settings dialog for this provider.

    Note

    In addition to capturing loopback traffic and unencrypted IPSEC messages, the Microsoft-PEF-WFP-MessageProvider minimizes other lower-level noise such as broadcast traffic, so that you can focus your analysis above the IP/Network Layer. Also note that messages below the Transport Layer are typically represented in the message stack as a WFPCapture and below that are events at the ETW level.

  3. If the Start With drop-down list in the New Session dialog does not indicate the Analysis Grid viewer, then click the drop-down list and select it.

  4. Click the Start button in the New Session dialog to start capturing data.

    Message Analyzer may immediately begin capturing and accumulating message data in the Analysis Grid viewer.

  5. While Message Analyzer is capturing data, launch a web browser and attempt to reproduce any conditions that are related to a particular HTTP issue you might be experiencing. For example, you might attempt to navigate to a poorly performing web server with your browser.

  6. Stop the trace at a suitable point by clicking the Stop button on the global Message Analyzer toolbar.

  7. Click the Add Columns button on the Analysis Grid viewer toolbar to display the Field Chooser Tool Window in focus in its default location, if it is not already displayed.

  8. Open the HTTP node in Field Chooser and navigate to the ContentType field in the HTTP Operation message hierarchy, right-click the field, and then select the Add As Column context menu item to add the ContentType column to the Analysis Grid viewer.

  9. Open the TCP node in Field Chooser and navigate to the Transport field in the Segment message hierarchy, right-click the field, and then select the Add As Column context menu item to add the Transport column to the Analysis Grid viewer.

  10. In a similar manner, add the ResponseTime field from the Global Annotations node in Field Chooser as a new column in the Analysis Grid viewer, so that you can determine how quickly the web server is responding to HTTP requests.

  11. Right-click the ContentType column in the Analysis Grid viewer and select the Group command from the context menu.

    The trace data is grouped according to the different content types associated with HTTP messages, so that you can examine the types of objects being passed to your web browser by the server.

Tip

To create a more focused analysis, you can limit the display to HTTP messages only by specifying an HTTP Viewpoint; you can do this by clicking the Viewpoint drop-down list on the Filtering toolbar and then selecting the HTTP item.

When you are done with assessing the data, click the x in the ContentType Group label above the Analysis Grid to remove the Group and return to the default view Layout.

  1. Sort the ResponseTime column in descending order and then Group this column to quickly expose the slowest server responses and associated messages in separate Groups for analysis.

  2. Remove the ResponseTime Group by clicking the x in the ResponseTime Group label above the Analysis Grid.

  3. Right-click the Transport column in the Analysis Grid viewer and select the Group command from the context menu.

    The trace data is grouped according to the different Transport types, such as TCP or UDP, so that you can examine the ports across which the most substantial traffic is transiting.

  4. Remove the Transport grouping in the previously indicated manner and then execute the Group command on the default Source or Destination column of the Analysis Grid viewer.

    The trace data is grouped according to Source or Destination, as appropriate, so that you can determine which IP addresses are carrying the most traffic. Note that you can obtain similar statistics by executing a Group command on the Network column, which you can add from the IP message hierarchy in the Field Chooser.

Tip

You can also nest groups by performing successive Group operations on multiple columns. For example, you might also add the Response.PayloadLength field from the HTTP message hierarchy in Field Chooser and then perform multiple groupings in succession on the ResponseTime, ContentType, PayloadLength, and Transport data columns so you can view the slowest server response times correlated with the content types that are associated with the highest payload values, along with the TCP ports that carried that information. Another grouped configuration you can try would be to Group the Source and Destination columns, in that order, to organize all the Destination traffic that is associated with each Source address, or vice versa.

  1. Remove all Group configurations to return to the original Analysis Grid viewer display and then execute the Group command on the DiagnosisTypes column in the Analysis Grid viewer.

    The trace data is grouped according to the different types of diagnosis messages, which includes Application, InsufficientData, Parsing, and Validation message types, so that you can immediately assess the types of errors that occurred in your trace. For more information about the meaning of these diagnosis message types, see the “Enum Values for DiagnosisType filters” table in the Diagnosis Category topic.

Perform Top-Level Summary Analysis

In the procedure that follows, you will use several viewing infrastructure components to accomplish simple data analysis tasks. For example, you will use the graphic chart visualizer components of the Protocol Dashboard viewer, which includes the Top Level Protocol Summary Bar element, Pie chart, Table grid, and the Timeline (Top Level Protocols Over Time) visualizer components, to view top-level protocol summary data that can reflect traffic volume levels for the message types in a trace, along with message activity across selected windows of time into which you can zoom. In the first part of the procedure, you will use the Microsoft-PEF-NDIS-PacketCapture provider in the Local Network Interfaces Trace Scenario to capture message data in a Live Trace Session. However, if you are running the Windows 8.1 or Windows Server 2012 R2 operating system, you will be using the Microsoft-Windows-NDIS-PacketCapture provider in this Trace Scenario.

This example also shows how to use the SMB Reads and Writes Bytes/Second, SMB File Stats, and SMB/SMB2 Service Performance view Layouts for Charts to expose file access activities and statistics. In this part of the procedure, you will start a new session with the Loopback and Unencrypted IPSEC Trace Scenario, in which you will use the Microsoft-PEF-WFP-MessageProvider to focus on statistical summaries of SMB/SMB2 file access operations messages above the IP/Network Layer.


More Information
To learn more about the Protocol Dashboard viewer, see the Protocol Dashboard topic.
To learn more about the SMB Layouts for the Chart viewer, see the subtopics in the File Sharing Category section.


To analyze top level summary data

  1. Perform steps 1 through 5 of the procedure To identify Transport and Network Layer messages with gradient-style Color Rules to start a Message Analyzer Live Trace Session with the Local Network Interfaces Trace Scenario.

  2. While Message Analyzer is capturing data, attempt to reproduce conditions that are related to any particular issue you might be trying to resolve, for example, a high volume of TCP traffic to a target computer.

  3. Stop the trace at a suitable point by clicking the Stop button on the global Message Analyzer toolbar.

  4. From the New Viewer drop-down list on the global Message Analyzer toolbar, click Charts (Deprecated) and then select the Protocol Dashboard item in the drop-down list.

  5. In the Protocol Dashboard viewer, observe the numerical and graphical presentation of top protocol activity in the trace by examining the Top Level Protocol Summary and Top Level Protocols Over Time visualizer components, which expose the relevant statistics.

    In the Top Level Protocol Summary Table and Bar element sections of the dashboard, you can observe message traffic volume that is sorted in a descending scale from highest to lowest. Note that an extraordinarily high traffic volume for a particular module can immediately expose the top bandwidth consumer, or exceptionally heavy TCP traffic might indicate that you have a large quantity of TCP retransmits or duplicate ACK messages in your trace.

    If you suspect there is an issue with a protocol or module that has particularly high traffic volume, you can double-click the Bar element or Pie chart segment representing the module in the Top Level Protocol Summary to display only those specific messages in a separate instance of the Analysis Grid viewer for further investigation. You can also adjust the time window slider controls of the Timeline visualizer component (Top Level Protocols Over Time) to zoom into specific messages in a particular time slot and then double-click a message node to display that traffic only in a separate instance of the Analysis Grid viewer for further investigation.

  6. Start another Live Trace Session with the Loopback and Unencrypted IPSEC Trace Scenario and capture data live with Message Analyzer while performing file access operations that have previously been problematic.

  7. Stop the trace at a suitable point by clicking the Stop button on the global Message Analyzer toolbar and then launch the SMB Reads and Writes Bytes/Second view Layout from the Chart drop-down in the New Viewer drop-down list on the global Message Analyzer toolbar.

    From this view Layout, you can obtain statistics that reflect the network bandwidth being consumed by the file access/sharing activities of the Server Message Block (SMB) protocols. See SMB Reads and Writes Bytes/Second for further details.

  8. Adjust the time window slider controls in the SMB Reads and Writes Bytes/Second viewer to zoom into specific messages in a particular time slot.

  9. Double-click a message node or time line in the SMB Reads and Writes Bytes/Second data viewer to display specific traffic in a new instance of the Analysis Grid viewer tab for further investigation.

Tip

You can also use the Field Chooser Tool Window to add an SMB FileName or SMB2 FileName column to the Analysis Grid viewer and then execute a Group command on the new column so that you can examine the SMB traffic that is associated with access to specific files.

  1. Optionally, select the SMB File Stats view Layout from the Chart drop-down in the New Viewer drop-down list. With this Layout, you can examine a summary of SMB file statistics in a Table visualizer component that includes the file name, access duration, total number of bytes for each file or folder access operation, and the data transmission rates, as described in SMB File Stats.

    You might also consider selecting the SMB/SMB2 Service Performance viewer Layout, also in the Chart drop-down, to examine statistics that expose how long first responses to SMB operations are taking (ResponseTime), possibly to expose slow server response issues; and how long it is taking for operations to complete (ElapsedTime), as a possible indication of network issues; as described in SMB/SMB2 Service Performance.

Perform Interactive Analysis with Data Viewers

The procedure that follows provides a simple example of how you might utilize the Analysis Grid and Protocol Dashboard viewers interactively to analyze captured message data. The example also indicates how you might user other data viewers interactively with the Analysis Grid viewer:


More Information
To learn more about data viewers and how they interact, see Data Viewer Concepts.


To analyze data through data viewer interaction

  1. Perform steps 1 through 4 of the procedure To identify Transport and Network Layer messages with gradient-style Color Rules, to start Message Analyzer, open the New Session dialog for Live Trace Session configuration, and select the Local Network Interfaces Trace Scenario.

  2. Click the Start button in the New Session dialog to begin capturing data in a Live Trace Session.

    The captured data begins to accumulate in the Analysis Grid viewer, assuming that you have not changed the default data viewer in the Default Profile pane on the Profiles tab of the global Options dialog that is accessible from the global Message Analyzer Tools menu.

  3. While Message Analyzer is capturing data, attempt to reproduce conditions that are related to any particular issue you are trying to isolate.

  4. Stop the trace at a suitable point by clicking the Stop button on the global Message Analyzer toolbar.

  5. From the New Viewer drop-down list on the global Message Analyzer toolbar, click Charts (Deprecated) and then select the Protocol Dashboard item in the drop-down list.

  6. In the Protocol Dashboard viewer, observe the relative distribution of captured message volumes in the Top Level Protocol Summary Bar element visualizer of the Protocol Dashboard, in an attempt to isolate message traffic that might be related to failures in a particular component, system, or service. Note that high message volumes can be a flag for underlying issues that may need further investigation, such as network issues and TCP retransmits.

  7. In the Bar element visualizer, double-click the graphic bar representing the message traffic you want to target, for example, a protocol that has a high volume of messages.

    A new instance of the Analysis Grid viewer opens and contains only the traffic that you targeted, for further analysis.

  8. Sort the DiagnosisTypes column in the Analysis Grid viewer to bubble up any errors that might have occurred in the target traffic.

  9. Perform a Group operation by right-clicking the DiagnosisTypes column in the Analysis Grid viewer and then selecting the Group command from the context menu that displays. The data is then organized into expandable group nodes that each contain different diagnosis message types. By expanding each node, you can view the messages that contain the diagnosis errors.

  10. Click the diagnosis error icons in the DiagnosisTypes column under the expanded group nodes to review the error message text. You might also examine the Summary column descriptions for these messages in the Analysis Grid viewer to discover any evidence of the underlying failures that are associated with the diagnosis errors that occurred. You can also review the values of the fields for selected messages in the Details Tool Window.


More Information
To learn more about the meaning of diagnosis message types, see the “Enum Values for DiagnosisType filters” table in the Diagnosis Category topic.


Apply Viewpoints to Trace Data

In the procedure that follows, you will apply HTTP and TCP Viewpoints so that you can view HTTP- or TCP-related traffic at top-level, without having to drill down into each message stack to expose these messages. In addition, you will alternately disable or enable Operations so that you can expose HTTP request and response messages — either in their original chronological order (by executing the Disable Operations Viewpoint), or encapsulated by Message Analyzer in top-level Operation rows (by executing the No Viewpoint command) respectively, in the Analysis Grid viewer.


More Information
To learn more about the concepts upon which this example procedure is based, see Applying and Managing Viewpoints.


To analyze data with applied Viewpoints

  1. From the Start menu, Start page, or task bar of your computer, click the Microsoft Message Analyzer icon to launch Message Analyzer.

    To ensure that you have access to all features, run Message Analyzer as Administrator.

  2. Click the Message Analyzer global File menu and then click New Session to display the New Session dialog.

  3. Under Add Data Source in the New Session dialog, click the Live Trace button to display the Live Trace tab along with the associated session configuration features that it contains in the New Session dialog.

  4. In the Network category of the Select Scenario drop-down list on the Live Trace tab, click the Loopback and Unencrypted IPSEC Trace Scenario.

    The Microsoft-PEF-WFP-MessageProvider information displays in the ETW Providers list on the Live Trace tab, which includes the provider Name, GUID, and a Configure link that opens the Advanced Settings dialog for this provider.

Note

In addition to capturing loopback traffic and unencrypted IPSEC messages, the Microsoft-PEF-WFP-MessageProvider minimizes other lower-level noise such as broadcast traffic, so that you can focus your analysis above the IP/Network Layer.

  1. If the Start With drop-down list in the New Session dialog does not indicate the Analysis Grid viewer, then click the drop-down list and select it.

  2. Click the Start button in the New Session dialog to start the trace and begin capturing data.

    The captured data begins to accumulate in the Analysis Grid viewer.

Note

You can start this type of trace immediately with default provider configuration settings, by selecting the Loopback and Unencrypted IPSEC Trace Scenario in the Favorite Scenarios list on the Message Analyzer Start Page. By starting a trace in this manner, you will not have access to provider configuration settings prior to capturing data.

  1. While Message Analyzer is capturing data, launch a web browser and attempt to reproduce any conditions that are related to a particular HTTP issue you are trying to isolate, for example, a slowly responding or non-responsive web server.

  2. Stop the trace at a suitable point by clicking the Stop button on the global Message Analyzer toolbar.

  3. In the Analysis Grid viewer, note that you have HTTP messages displaying as top-level operation message rows, as signified by messages with a blue-cubed icon to the left of the message number, along with some TCP messages at top-level and others hidden within the message stack. You may also have HTTP fragments hidden within Analysis Grid viewer expansion nodes. This configuration of displayed messages is typical of the results returned by the Microsoft-PEF-WFP-MessageProvider, which focuses on Transport Layer messages and above.

  4. Click the Viewpoints drop-down list on the Filtering toolbar and select the HTTP item from the list.

    All HTTP messages are driven to top-level message rows in the Analysis Grid viewer, which can also include any fragments that existed in the origins tree (message stack). In this view configuration, you can focus on HTTP messages without the encumbrance of other message types in display. However, because associated HTTP request and response messages are still grouped as Operations to provide context, there is a chronological displacement of response messages in this configuration that you can resolve by disabling the Operations. See the Important note below.

  5. Click the Disable Operations item in the Viewpoints drop-down list on the Filtering toolbar just below the Analysis Grid viewer tab.

    Note that the HTTP request and response message pairs that were formerly grouped under Operation nodes are now displayed in chronological order in the Analysis Grid viewer. This view should be familiar to Network Monitor users and should enable them to work with the messages in the manner with which they are accustomed. However, for quick analysis of HTTP request and response pairs, it is more expedient to view the data encapsulated in the default Operation node format to see the information at top-level. At this level, you can still use typical data analysis tools such as sorting, grouping, filtering, and so on.

    Also note that data values that are important to HTTP analysis include ResponseTime and ElapsedTime, as described in the Important note below.

  6. To return to the default display, click the No Viewpoint item in the Viewpoints drop-down list on the Filtering toolbar to return to the default viewpoint. This is the presentation format in which the Analysis Grid viewer normally displays.

  7. Next, isolate TCP messages at top-level in the trace by selecting the TCP item in the Viewpoint drop-down list.

    All TCP messages display in top-level message rows in the Analysis Grid viewer. Note that you will not see any Operation message nodes in this view because the TCP Viewpoint filters out everything above it.

  8. From the TCP Viewpoint, use your typical data analysis tools such as sorting, grouping, filtering, pattern matching, and annotating, in addition to viewing message Details, Message Stack, and Diagnostics Tool Window data, to analyze the information. You might also apply the built-in TCP Deep Packet Analysis with Absolute Sequence Number and Grouping Layout from the Layout drop-down list on the Analysis Grid toolbar, so you can focus on important TCP field data.

Important

Pairing up request and response messages in Operation nodes for protocols that typically use request/response pairs such as HTTP, DNS, and SMB2, provides immediate access to response messages rather than having to search through potentially hundreds or even thousands of messages to find them. Another advantage of this configuration is that you can readily measure and correlate important values such as ResponseTime and ElapsedTime, which specify how long it took for the first server response and how long it took to receive all message fragments and complete the Operation, respectively. High values for these times can provide an indication of a poorly responding server in the first case and network latency issues in the second. The ElapsedTime is displayed by default in the Analysis Grid viewer column layout; however, you must add the ResponseTime column by right-clicking it under Global Annotations in the Field Chooser Tool Window and then selecting the Add as Column command.

More Information
To learn more about average response time for Operations, see Average Response Time for Operations.
To learn more about average elapsed time for Operations, see Average Elapsed Time for Operations.

Apply a Time Filter to Trace Results

In the procedure that follows, you will start a Data Retrieval Session and load data from a saved trace or log file. You will then apply a Time Filter to the loaded message collection so that you can temporarily focus on analyzing messages in a specified window of time. You will also verify that you can toggle back and forth between the time-filtered data and your original data, as your analysis might require.


More Information
To learn more about the concepts upon which this example procedure is based, including the benefits of using a Time Filter, see Applying a Time Filter to Session Results.


To apply a Time Filter to a set of trace results

  1. From the Start menu, Start page, or task bar of your computer, click the Microsoft Message Analyzer icon to launch Message Analyzer.

    To ensure that you have access to all features, run Message Analyzer as an Administrator.

  2. Click the global Message Analyzer File menu and then click New Session to display the New Session dialog.

  3. Under Add Data Source in the New Session dialog, click the Files button to display the Files tab along with the associated session configuration features that it contains in the New Session dialog for a Data Retrieval Session.

  4. On the Files tab of the New Session dialog, click the Add Files button on the Files tab toolbar to launch the Open dialog, select a large trace or log file containing data that you want to view in a specific time window, and then click Open.

The name of the trace or log file appears in the files list.

  1. If you loaded a *.log file, you should choose an applicable configuration file from the Text Log Configuration drop-down list just below the toolbar of the Files tab for your log, to enable full parsing of messages.

Important

A built-in OPN configuration file is required to parse a text log. Message Analyzer provides several configuration files by default, including Cluster, IIS, SambaSys, and Netlogon .log files, among many others. If the type of text log you want to parse is not included in the previously specified list, you will need to create your own custom OPN configuration file to parse your log, as described in Addendum 1: Configuration Requirements for Parsing Custom Text Logs.

  1. Confirm that the Analysis Grid viewer is selected in the Start With drop-down list and then click the Start button in the New Session dialog to start loading the message data.

Important

Do not configure a Time Filter in the Data Retrieval Session configuration, given that you will be doing this only after the data is loaded into Message Analyzer. You might do this in a different scenario where you want to improve performance by limiting the amount of input messages from a high volume data file, as described in Considering Performance vs. Usability Factors for Time Filter Application, but not in this particular example.

  1. After the message data is loaded and displayed in the Analysis Grid viewer, click the Add Time Filter item from the Add Filter drop-down list on the Filtering toolbar to open the Time Filter panel.

  2. In the Time Filter panel, use the Start Time and End Time slider controls to configure a window of time in which you want to view data. As you do this, you will see the Start Time and End Time values change.

  3. When you are done with Time Filter configuration, click the Apply button in the Time Filter panel to filter the message data according to the time window that you specified.

    The number of messages displaying in the default data viewer is reduced in accordance with the specified Time Filter configuration, thus enabling you to perform analysis on a focused data set.

    Note that the number of messages that display are indicated next to the Available label on the Message Analyzer status bar at the bottom of the UI.

  4. To remove the time filtering configuration that you applied, click the Remove button in the Time Filter panel to return to your original data.

  5. To reapply the same time filtering configuration, click the Apply button again.

    Note that you can toggle the application and removal of a Time Filter as many times as your analysis requires. You can also apply view Filters, Viewpoints, and Viewpoint Filters to the time filtered results to further focus the results you want to analyze.

Note

Time Filters do not persist across sessions or even across viewers of the same session, which means that you will need to create a new Time Filter configuration for every session or viewer where you want to apply time window filtering. Also note that you can save any data set to which you have applied a Time Filter by clicking Save As on the global Message Analyzer File menu and specifying the Filtered Messages for Analysis Grid view option to perform the save with the Save/Export Session dialog.

Drive Analysis Grid Viewer and Tool Window Interactions

In the procedure that follows, you will run a trace and display data in the Analysis Grid viewer. Thereafter, through message selection in the Analysis Grid viewer, or message and field selection in various Tool Windows, the procedure will demonstrate how to interactively drive the display of data in these viewing components to facilitate rapid assessment of message details, which include field values and types, hexadecimal or binary data, diagnosis message types and details, message stack configurations, and so on. This procedure assumes that certain tool windows you will be working with are not currently displayed in the Message Analyzer analysis surface. If they are already displayed, please ignore the steps that specifically require you to display them.


More Information
To learn more about how to position Message Analyzer data viewers and Tool Windows for enhanced data analysis, see Working with Message Analyzer Window Layouts.


To drive interaction between the Analysis Grid viewer and tool windows

  1. Perform steps 1 through 7 of the procedure To identify Transport and Network Layer messages with gradient-style Color Rules to start and stop a new Message Analyzer Live Trace Session that uses the Local Network Interfaces Trace Scenario.

  2. Click the global Message Analyzer Tools menu, click the Windows item, and then select Diagnostics in the submenu that appears, to display the Diagnostics Tool Window in its default docking location.

Note

The Diagnostics window is a preview feature that will not be included in the Windows submenu unless you have first selected it on the Features tab of the Options dialog. This dialog is also accessible from the global Message Analyzer Tools menu. Note that a Message Analyzer restart is required after this selection.

  1. Click the global Tools menu again, click the Windows item, click the Message Stack item, and then select Message Stack 1 in the submenu to display the Message Stack 1 Tool Window in its default docking location.

  2. Ensure that the Message Stack 1 window is in focus (click its tab) and then select any message in the Analysis Grid viewer.

    Observe that message selection in the Analysis Grid viewer drives message selection in the Message Stack 1 window and message details in the Details Tool Window.

  3. Click the Diagnostics window tab in its default docking location to bring it into focus and then select one or more diagnosis message types in the Diagnostics grid.

    Observe that message selection in the Diagnostics window drives selection of one or more top-level messages in the Analysis Grid viewer and corresponding message details in the Details window.

Note

You should be aware that even though top-level messages are highlighted, the actual message that contains a diagnosis error might be at a lower layer in the origins tree (message stack). You can determine this by expanding message nodes in the Analysis Grid viewer under the highlighted top-level message/s. In addition, note that the Diagnostics window data columns provide descriptions of all diagnosis messages in the current trace results, so you do not have to drill down into the origins tree through node expansion to see them.

  1. Click the Message Data 1 Tool Window tab in its default docking location to bring it into focus and then select any message in the Analysis Grid viewer.

    Observe that message selection in the Analysis Grid viewer drives the display of hexadecimal, binary, or ASCII data selection in the Message Data 1 window and message details in the Details window.

Note

If you select the top-level message in any Operation row, it does not display any data in any Message Data window. Rather, you must expand the Operation node in the Analysis Grid viewer and select one of the nested request or response messages that it contains to display highlighted hexadecimal data.

  1. Select any message in the Analysis Grid viewer and then select a field Name in the Details window.

    Observe that field selection in the Details window drives the display of a hexadecimal, binary, or ASCII field value in any Message Data window and a field value in the Field Data Tool Window as well, provided that the field you selected in Details had a field value.

Create an Alias for a Data Field Value

In the procedure that follows, you will perform a live trace and display the results data in the Analysis Grid viewer. You will then create an Alias for an IPv6 address and name it with a string value of “MyComputer”. You will then use the new Alias in a Filter Expression that you apply to the trace.


More Information
To learn more about the concepts upon which this example procedure is based, see Using and Managing Message Analyzer Aliases.


To create a field value Alias

  1. Perform steps 1 through 7 of the procedure To identify Transport and Network Layer messages with gradient-style Color Rules to start and stop a new Message Analyzer Live Trace Session that uses the Local Network Interfaces Trace Scenario.

  2. In the Destination column of the Analysis Grid viewer, right-click an IPv6 address for the local computer and select the Create Alias for ‘Destination’… item in the context menu that displays. The Alias Editor dialog displays, in which you can specify an Alias name, Description, and Category.

    If you do not know the IPv6 address of the local computer, run IPConfig /All at the command line to determine it.

  3. In the Alias text box of the Alias Editor dialog, specify a friendly name such as “MyComputer”, or specify another name that is appropriate for your environment.

  4. In the Description text box of the Alias Editor dialog, enter a description that identifies the purpose of the Alias, for future reference and for identification if you intend to share the Alias with other users.

    The Description text will display in a tool tip when you hover over the Alias name in the Aliases drop-down list with your mouse, or when you hover over the Alias name in the Manage Alias dialog.

  5. In the Category combo box of the Alias Editor dialog, either select an existing Category or specify a new one, for example “IPv6 Addresses”.

    Any new Category that you specify appears as a subcategory under the top-level My Items category and will contain the new Alias after you Save it.

  6. In the Alias Editor dialog, ensure that the Auto Refresh Views check box is selected if you want Message Analyzer to immediately perform a refresh of all data viewers that will be impacted by application of the new Alias.

  7. In the Alias Editor dialog, click the Save button to save your new Alias.

    All data viewers, including the Analysis Grid viewer and any Chart viewer Layout that is displayed, are updated to reflect application of the new Alias, providing that the Auto Refresh Views check box was selected when you saved the Alias. If this is the case, observe that the IPv6 address of the local computer is now identified in the Source and Destination address columns of the Analysis Grid viewer as “MyComputer”.

    Also verify that the new Alias appears in the Aliases drop-down list, which is accessible from the global Message Analyzer Tools menu. In a similar manner, the Alias should also appear in the Manage Alias dialog.

  8. In the text box on the default Filter panel of the Filtering toolbar, enter the following text to create a Filter Expression that uses your new Alias:

    *Source == “MyComputer”

Note

If a Filter panel on the Filtering toolbar is not currently displayed, click the Add Filter drop-down list just below the Analysis Grid viewer tab and select the Add Filter item in the list.

  1. Click the Apply button on the Filter panel of the Filtering toolbar and observe that the specified filter removes all traffic except the messages in which “MyComputer” represents either the Source or Destination computer IPv6 address.

Note

If you want to save the filter you created in this procedure, select the New Filter item in the Library drop-down on the Filter panel of the Filtering toolbar and provide a Name, Description, and Category for the filter in the Edit Filter dialog that appears before you Save it. Note that the Edit Filter dialog automatically captures the Filter Expression text that you specified in the Filter panel text box.

Create a Union of Two Data Fields

In the procedure that follows, you will load data into Message Analyzer from two saved files that contain related data that was captured in a common environment and within the same time frame; one from a log file and the other from a former live trace. The procedure specifies files that contain messages from SMB operations that have identical value data for certain fields, but which are named differently. After you load your data files into Message Analyzer, the data should display in the Analysis Grid viewer and in an interlaced fashion. Thereafter, you will create a Union that combines two fields of equal value but with different names into a single field with a new name, to simplify your data analysis processes with Message Analyzer.

Important

Because creating a working Union in your Message Analyzer installation depends on combining fields that are specific to your environment, the procedure that follows uses hypothetical field names, such as Command.smb_cmd and Command. Therefore, when using this procedure to create a working Union, you should substitute actual field names that are contained in actual data files that are specific to your environment. In addition, you have the option to specify any Union name that is appropriate for your needs.


More Information
To learn more about the concepts upon which this example procedure is based, see Configuring and Managing Message Analyzer Unions.


  1. From the Start menu, Start page, or task bar of your computer, click the Microsoft Message Analyzer icon to launch Message Analyzer.

    To obtain access to all features, be sure to run Message Analyzer as an Administrator.

  2. From the Message Analyzer Start Page, click the New Session button to display the New Session dialog.

  3. Under Add Data Source in the New Session dialog, click the Files button to display the Files tab along with the associated session configuration features that it contains in the New Session dialog for a Data Retrieval Session.

  4. On the Files tab of the New Session dialog, click the Add Files button on the Files tab toolbar to launch the Open dialog, select the trace and log files that contain the data fields for which you will create a Union, and then click Open.

Tip

You will need to know in advance the field names from the messages in your input files for which you will be creating a new Union. As an example, the built-in SMBTID union that you can access from the global Message Analyzer Tools menu creates a Union of the following three fields that are accessible in Field Chooser Tool Window under the SMB, SMB2, and SambaSysLog nodes, respectively:

SMB.SmbHeader.Tid
SMB2.SMB2Request.Header.TreeId
SambaSysLog.smb_command.command.smb_tid - note that the SambaSysLog node and this field will only exist in Field Chooser if a Samba *.log file is loaded into Message Analyzer with a SambaSysLog configuration file specified in the New Session dialog for a Data Retrieval Session. This could apply if you are loading a *.log file into Message Analyzer.

Whenever you create a new Union, you will need to use the Field Chooser to locate and add the fields that are to comprise the Union, as indicated ahead in this procedure.

The name of the trace and log files appear in the files list.

  1. Observe the current Start With drop-down list selection in the New Session dialog; if it is not the Analysis Grid viewer, click the drop-down list and select the Analysis Grid item.

  2. Click the Start button in the New Session dialog and observe that the messages loaded into Message Analyzer from the log and trace files display in a chronological interlaced fashion in the Analysis Grid viewer, with a column for each differently named field of interest that displays similar values in each corresponding column.

  3. Click the Unions button on the global Message Analyzer toolbar to display the Edit Union dialog.

  4. In the Edit Union dialog, perform the following actions:

    • In the Name text box, specify a name for the Union. Be sure to enter a name that is meaningful in your environment. In this example, the hypothetical Union name is SMBCommand2.

    • In the Category combo-box, either select an existing Category or type a new one.

    • To add the fields you want to combine in the Union, click the Add button to display the Field Chooser Tool Window, in which you can locate the field names. Note that you can add only one field at a time with the Field Chooser. In this example, the hypothetical field names are Command.smb_cmd and Command.

      As you add fields, you should notice the Type label displaying the most appropriate data type for the combined fields, as calculated by Message Analyzer; see Creating Unions for more information.

    • When you are finished configuring the Union, click the Save button in the Edit Union dialog.

      The new Union is added to the root Unions node in the Field Chooser window.

  5. Open the Field Chooser window by clicking the Add Columns button on the Analysis Grid viewer toolbar.

  6. Expand the root Unions node in the Field Chooser and then double-click the name of the new Union to add it as a new data column in the Analysis Grid viewer column layout.

    Observe that the new <unionName> column in the Analysis Grid viewer correlates the data field values for the disparate field names that you specified in the Union you created. Note that you can remove the disparate field columns from the Analysis Grid viewer by selecting the Remove command that displays as a context menu item when you right-click the corresponding column header for each field. When the original field columns are removed from the Analysis Grid viewer, the Union name column will continue to correlate the values for the underlying data fields contained in the Union.

See Also

Configuring and Managing Message Analyzer Unions