Monitor historical trends across a site
The Trends view accesses historical trend information of each site for the following parameters:
- connection failures
- machine failures
- logon performance
- load evaluation
- capacity management
- machine usage
- resource utilization
To locate this information, click the Trends menu.
The zoom-in drill down feature lets you navigate through trend charts by zooming in on a time period (clicking a data point in the graph) and drilling down to see the details associated with the trend. This feature enables you to better understand the details of who or what has been affected by the trends being displayed.
To change the default scope of each graph, apply a different filter to the data.
- Sessions, failures, and logon performance trend information are available as graphs and tables when the time period is set to Last month (Ending now) or shorter. When the time period is chosen as Last month with a custom ending date or as Last year, the trend information is available as graphs but not as tables.
- Citrix Virtual Apps and Desktops service supports historical data retention only for 90 days. Hence, one-year trends and reports in Monitor show the last 90 days of data.
View trends for sessions: From the Sessions tab, select the delivery group and time period to view more detailed information about the concurrent session count.
The Session Auto Reconnect column displays the number of auto reconnects in a session. Auto reconnect is enabled when the Session Reliability or the Auto Client Reconnect policies are in effect. When there is a network interruption on the endpoint, the following policies come into effect:
Session reliability comes into effect (by default for 3 minutes) where the Citrix Receiver or Citrix Workspace app tries to connect to the VDA.
Auto client reconnect comes into effect between 3 and 5 minutes where the client tries to connect to the VDA.
Both these reconnects are captured and displayed to the user. This information can take a maximum time of 5 minutes to appear on the Director UI after the reconnect occurs.
The auto reconnect information helps you view and troubleshoot network connections having interruptions, and to analyze networks having a seamless experience. You can view the number of reconnects for a specific delivery group or time period selected in the Filters.
A drilldown provides additional information like Session Reliability or Auto Client Reconnect, time stamps, Endpoint IP, and Endpoint Name of the machine where Workspace app is installed.
By default, logs are sorted by the event time stamps in descending order. This feature is available for Citrix Workspace app for Windows, Citrix Workspace app for Mac, Citrix Receiver for Windows, and Citrix Receiver for Mac. This feature requires VDAs 1906 or later.
Sometimes, the auto reconnect data might not appear in Monitor for the following reasons:
Workspace app is not sending auto reconnect data to VDA.
VDA is not sending data to monitor service.
Sometimes, the client IP address might not be obtained correctly if certain Citrix Gateway policies are set.
View trends for connection failures: From the Failures tab, select the connection, machine type, failure type, delivery group, and time period to view a graph containing more detailed information about the user connection failures across your site.
View trends for machine failures: From the Single session OS Machine Failures tab or Multi-session OS Machines tab, select the failure type, delivery group, and time period to view a graph containing more detailed information about the machine failures across your site.
View trends for logon performance: From the Logon Performance tab, select the delivery group and time period to view a graph containing more detailed information about the duration of user logon times across your site and whether the number of logons affects the performance. This view also shows the average duration of the logon phases, such as brokering duration and VM start time. This data is specifically for user logons and does not include users trying to reconnect from disconnected sessions. The table below the graph shows Logon Duration by User Session. You can choose the columns to display and sort the report by any of the columns. For more information, see Diagnose user logon issues.
View trends for load evaluation: From the Load Evaluator Index tab, view a graph containing more detailed information about the load that is distributed among Multi-session OS machines. The filter options for this graph include the delivery group or Multi-session OS machine in a delivery group, Multi-session OS machine (available only if Multi-session OS machine in a delivery group was selected), and range. The Load Evaluator Index is displayed as percentages of Total CPU, Memory, Disk or Sessions and is shown in comparison with the number of connected users in the last interval.
View hosted applications usage: From the Capacity Management tab, select the Hosted Applications Usage tab, select the delivery group and time period to view a graph displaying peak concurrent usage and a table displaying application based usage. From the Application Based Usage table, you can choose a specific application to see details and a list of users who are using, or have used, the application. You can see the predicted peak concurrent application instances values chosen future time period with Application instance prediction. For more information, see the Application instance prediction section.
View single and multi-session OS usage: The Trends view shows the usage of Single session OS by site and by delivery group. When you select site, usage is shown per delivery group. When you select delivery group, usage is shown per User. The Trends view also shows the usage of Multi-session OS by site, by delivery group, and by Machine. When you select site, usage is shown per delivery group. When you select delivery group, usage is shown per Machine and per User. When Machine is selected usage is shown per User.
View virtual machine usage: From the Machine Usage tab, select Single session OS Machines or Multi-session OS Machines to obtain a real-time view of your VM usage. The page displays the number of Autoscale enabled Multi-session and Single session OS machines that are powered on for a selected delivery group and time period. Also available is the estimated savings achieved by enabling Autoscale in the selected delivery group, this percentage is calculated using the per machine costs.
The usage trends of Autoscale enabled machines indicate the actual usage of the machines, enabling you to quickly assess your site’s capacity needs.
- Single session OS availability - displays the current state of Single session OS machines (VDIs) by availability for the entire site or a specific delivery group.
- Multi-session OS availability - displays the current state of Multi-session OS machines by availability for the entire site or a specific delivery group.
The grid below the chart displays the delivery group based machine usage data in real-time. The data includes machine availability of all machines independent of Autoscale enablement. The number of machines displayed in the Available Counter column in the grid includes machines in maintenance mode.
The monitoring data consolidation depends on the time period you select.
- Monitoring data for the one day and one week time periods is consolidated per hour.
- Monitoring data for the one month time period is consolidated per day.
The machine status is read at the time of consolidation and any changes during the period in between is not considered. For the consolidation period, refer to the Monitor API documentation.
For more information on monitoring Autoscale enabled machines see the Autoscale article.
View resource utilization: From the Resource Utilization tab, select Single session OS Machines or Multi-session OS Machines to obtain insight into historical trends data for CPU and memory usage, and IOPS and disk latency for each VDI machine for better capacity planning. This feature requires VDAs version 7.11 or later. Graphs show data for average CPU, average memory, average IOPS, disk latency, and peak concurrent sessions. You can drill down to the machine, and view data and charts for the top 10 processes consuming CPU. Filter by delivery group and Time period. CPU, memory usage, and peak concurrent sessions graphs are available for the last 2 hours, 24 hours, 7 days, month, and year. The average IOPS and disk latency graphs are available for the last 24 hours, month, and year.
- The Monitoring policy setting, Enable Process Monitoring, must be set to "Allowed" to collect and display data in the Top 10 Processes table on the Historic Machine Utilization page. The policy is set to "Prohibited" by default. All resource utilization data is collected by default. This can be disabled using the Enable Resource Monitoring policy setting. The table below the graphs shows the resource utilization data per machine.
- Average IOPS shows the daily averages. Peak IOPS is calculated as the highest of the IOPS averages for the selected time range. (An IOPS average is the hourly average of IOPS collected during the hour on the VDA).
View application failures: The Application Failures tab displays failures associated with the published applications on the VDAs.
This feature requires VDAs version 7.15 or later. Single session OS VDAs running Windows Vista and later, and Multi-session OS VDAs running Windows Server 2008 and later are supported. For more information, see Historical application failure monitoring.
By default, only application faults from Multi-session OS VDAs are displayed. You can set the monitoring of application failures by using Monitoring policies. For more information, see Monitoring policy settings.
View application probe results: The Application Probe Results tab displays the results of probe for applications that have been configured for probing in the Configuration page. Here, the stage of launch during which the application launch failure occurred is recorded.
This feature requires VDAs version 7.18 or later. For more information see Application probing.
Create customized reports: The Custom Reports tab provides a user interface for generating Custom Reports containing real-time and historical data from the Monitoring database in tabular format.
From the list of previously saved Custom Report queries, you can click Run and download to export the report in CSV format, click Copy OData to copy and share the corresponding OData query, or click Edit to edit the query. You can create a Custom Report query based on machines, connections, sessions, or application instances. Specify filter conditions based on fields such as machine, delivery group, or time period. Specify extra columns required in your Custom Report. Preview displays a sample of the report data. Saving the Custom Report query adds it to the list of saved queries.
You can create a Custom Report query based on a copied OData query. To do this, select the OData Query option and paste the copied OData query. You can save the resultant query for execution later.
The column names in Preview and Export report generated using OData queries are not localized, but appear in English.
The flag icons on the graph indicate significant events or actions for that specific time range. Hover the mouse over the flag and click to list events or actions.
- HDX connection logon data is not collected for VDAs earlier than 7. For earlier VDAs, the chart data is displayed as 0.
- Delivery groups deleted in Citrix Studio are available for selection in the Trends filters until data related to them are groomed out. Selecting a deleted delivery group displays graphs for available data until retention. However, the tables don’t show data.
- Moving a machine containing active sessions from one delivery group to another causes the Resource Utilization and Load Evaluator Index tables of the new delivery group to display metrics consolidated from the old and new delivery groups.
Application instance prediction
Predictive analytics gives you the ability to predict future resource usage. This feature is especially useful for administrators to organize required resources and licenses on each resource.
The first predictive analysis feature, Application instance prediction predicts the number of hosted application instances likely to be launched per site or delivery group over time.
Application instance prediction is available in the Trends > Capacity Management tab that displays the hosted application usage for the chosen time period. The historical graph contains the peak concurrent application instances values plotted for the chosen period.
To get the predicted graph, select the Predict check box. A dotted line prediction graph is displayed as an extension of the historical graph. The predicted peak concurrent application instances values are plotted with the time line extended into the future for the chosen time period.
You can predict the application instances for time periods of the next 7 days, 1 month, or one year. Custom ending dates are not supported.
Prediction is done using machine learning algorithms that are based on data models created with existing historical data. The predictions are therefore as accurate as the quality of the existing data.
The accuracy of prediction is indicated by the tolerance level that is displayed as a tool tip over the predicted graph. It indicates the amount of possible variation of the actual values from the predicted values.
The tolerance level can be high if either the available data does not follow a regular pattern or is missing for certain periods or is insufficient.
Prediction for a year captures the monthly and quarterly patterns coupled with the overall trend for the year. Similarly, monthly prediction captures the daily and weekly patterns along with weekly trends such as reduced activity over weekends.
Sufficient historical data must be available for prediction as follows:
- 14 days data for 7 days’ prediction
- 35 days data for one month’s prediction
- 84 days data for one year’s prediction
You can export only the historical graph, but not the predicted graph.
For a short demonstration of the Application instance prediction feature, see the following video.