Introduction from autor
Recently I was invastigating some logon problems. In my Citrix environment I have EdgeSight monitoring. So I have opened “Session Startup Duration” report and started reading. But there are many abbreviations which meaning I forget. So I’ve tried to find appropriate document. In the past I’ve found it in EdgeSight Reporting Wiki pages. I had luck, because those page was cached by google. All links was broken. Now, the whole ES documentation is combine in one PDF document. You can download it from Citrix or from my blog.
After that I’ve decided to put here whole page that describe informations about Session Startup Duration. Here is the original Citrix document (in the past you could found it here: Citrix Developer Network > EdgeSight > Citrix EdgeSight Reporting Wiki (Beta) > Contents > Reports (Browse) > EdgeSight Report List> EdgeSight 5.2 Report List > Session Startup Duration Details).
Original Citrix document
The Session Startup Duration Details report can provide you with information to help you troubleshoot long session start times sessions. The report provides metrics for how long the server and client take to perform their startup tasks.
This article provides an overview of how to interpret this report. It is designed to be used in conjunction with the article, Improving Slow Session Start Times.
A screen capture of the Session Startup Duration Details report from EdgeSight 5.0, SP2.
How satisfied users are with their Citrix environment is often directly tied to how long it takes for sessions to start. However, investigating long session start times can be complex since XenApp performs many different tasks while sessions start.
To isolate the cause of long session start-times, you can use the Session Startup Duration Details report that is described in this article.
This report provides metrics for both the client and server startup durations.
This section provides detail about the meaning of the metrics in this report. The detail is based on the information in the whitepaper, Improving visibility to the user login experience with Citrix EdgeSight. However, the tables below contain additional information and suggested actions for remedying issues. In other words, the tables below are worth reading even if you are familiar with the article.
These metrics summarize the server and client side of the startup transaction.
|Session Start-up Server Duration (SSD)||This is the high-level server-side connection start-up metric that encompasses the time XenApp takes to perform the entire start-up operation.
When an application starts in a shared session, this metric is normally much smaller than when starting a new session, which involves potentially high-cost tasks such as profile loading and login script execution.
|When this metric is high, it indicates that there is a server-side issue increasing session start times.|
|Client Start-up Duration (CSD)||When this metric is high, it indicates a client-side issue that is causing long start times.||Review subsequent metrics in this table to determine the probable root cause of the issue.|
These metrics measure data on the server-side of the startup transaction.
|Credentials Authentication Server Duration (CASD)||The time the application server spends authenticating the user’s credentials against the authentication provider, which may be Kerberos, Active Directory, or anSecurity Support Provider Interface (SSPI).|
|Credentials Obtention Network Server Duration (CONSD)||The time spent by the server performing network operations to obtain credentials for the user.||This only applies to a Security Support Provider Interface login (a form of pass-through authentication where the client device is a member of the same domain as the server and Kerberos tickets are passed in place of manually entered credentials).|
|Credentials Obtention Server Duration (COSD)||The time taken for the server to obtain the user credentials.
Because this metric may be artificially inflated if a user fails to provide credentials in a timely manner, it is not included in the Session Start-up Server Duration (SSD).
|This time is only likely to be a significant if manual login is being used and the server-side credentials dialog is displayed (or if a legal notice is displayed before login commences).|
|Program Neighborhood Credentials Obtention Server Duration (PNCOSD)||The time needed for the server to cause the Program Neighborhood instance running on the client to obtain the user credentials.||Like the COSD metric, this metric is not included in the Session Startup Server Duration (SSD) because it may be artificially inflated if a user does not enter credentials efficiently.|
|Profile Load Server Duration (PLSD)||The time required for the server to load the user’s profile.||
If this metric is high, consider your Terminal Services profile configuration.
Citrix Consulting has found that when customers have logon times greater than 20 seconds, in most cases, this can be attributed to poor profile and policy design. Roaming profile size and location contribute to slow session starts. When a user logs onto a session where Terminal Services roaming profiles and home folders are enabled, the roaming profile contents and access to that folder are mapped during logon, which takes additional resources. In some cases, this can consume significant amounts of the CPU usage. Consider using the Terminal Services home folders with redirected personal folders to mitigate this problem. In general, consider using Citrix Profile management to manage user profiles in Citrix environments. This tool also provides logging capabilities to help isolate profile issues.
For Terminal Server profiles best practices, see the whitepaper, Citrix User Profile Manager, User Profile Best Practices for XenApp 5_ _(CTX120285), Top 10 Items Found by Citrix Consulting on Assessments.
If you are using Citrix profile management and have slow logon times, check to see if your antivirus software is blocking the Citrix profile management tool. This is described further in the “Profile Streaming and Enterprise Antivirus Product”topic in Citrix eDocs.
|Login Script Execution Server Duration (LSESD)||The time the server needs to run the user’s login scripts.||Consider if you can streamline this user or group’s login scripts. Consider if you can optimize any application compatibility scripts or use environment variables instead.|
|Printer Creation Server Duration (PCSD)||The time required for the server to synchronously map the user’s client printers. If the configuration is set such that printer creation is performed asynchronously, no value is recorded for PCSD as it is does not impact completion of the session start-up.||Excessive time spenting mapping printers is often the result of the printer autocreation policy settings. The number of printers added locally on the users’ client devices and your printing configuration can directly affect your session start times. When a session starts, XenApp has to create every locally mapped printer on the client device.
Consider reconfiguring your printing policies to reduce the number of printers that get created – especially if users have a lot of local printers. To do this, in Presentation Server 4.5 and XenApp 5.0, edit the Printer Autocreation policy.
|Drive Mapping Server Duration (DMSD)||The time needed for the server to map the user’s client drives, devices and ports.||Make sure that, when possible, your base policies include settings to disable unused virtual channels, such as audio or COM port mapping, to optimize the ICA protocol and improve overall session performance.|
|Session Creation Server Duration(SCSD)||The time the server spends creating the session. This should not be confused with the overall SSD.||The session start times issue occurs between the when client connection is established and authentication begins.|
This table provides a list of the Client Startup metrics that appear in this report. If you do not have a client configured that meets the minimum prerequisites for gathering EUEM data, this row will appear blank. Likewise, if you are using EdgeSight for Load Test (ESLT) to simulate user transactions, the client row will be blank. While ESLT can simulate client behavior, it isn’t an actual ICA-based client. As a result, while ESLT can create transactions that reveal server startup times, it cannot generate data for client startup times.
For detailed information about which metrics are gathered for which clients, see the appendices in Improving visibility to the user login experience with Citrix EdgeSight.
|Metric Abbreviation||Meaning||Suggested Actions|
|Application Enumeration Client Duration (AECD)||Application enumeration is one of the issues slowing down session start times.||Consider if the cause is an overloaded XML Broker or Web Interface server.|
|BUCC (Backup URL Client Count)||If this metric has a value higher than 1, it indicates the Web Interface server is unavailable and the XenApp Plugin (formerly known as Program Neighborhood Agent) is attempting to connect to back-up Web Interface servers to launch the application.||
|Configuration File Download Client Duration (CFDCD)||The time it takes to get the configuration file from the XML server.|
|Credentials Obtention Client Duration (COCD)||The time it takes to obtain user credentials. This is a good metric to subtract from other client-side metrics.
Note: COCD is only measured when the credentials are entered manually by the user. Because this metric may be artificially inflated if a user fails to provide credentials in a timely manner, it is subtracted from the Startup Client Duration (SCD). This consideration is especially important if the metric is to be used for threshold alerting.
|Subtract this metric from other client-side metrics.|
|ICA File Download Duration (IFDCD)||The time it takes for the plugin (client) to download the ICA file from the server. The overall process is:
||If IFDCD is slow (but LPWD is normal), the server-side processing of the launch was successful, but there were communication issues between the client device and the Web server.
Often, this results from network trouble between the two machines, so investigate potential network issues first.
|Launch Page Web Server Duration (LPWD)||Review the information for IFDCD. The LPWD metric is only used when Web Interface is the application launch mechanism. If LPWD is slow, there is a bottleneck on the Web Interface server.||Possible causes include:
|Name Resolution Client Duration (NRCD)||This metric is collected when a client device directly queries the XML Broker to retrieve published application information stored in IMA (for example, when using Program Neighborhood or a Custom ICA Connection). NRCD is only gathered for new sessions since session sharing occurs during startup if a session already exists.||When this metric is high, it indicates the XML Broker is taking a lot of time to resolve the name of a published application to an IP address. Possible causes include a problem on the client, issues with the XML Broker, such as the XML Broker being overloaded, a problem with the network link between the two, or a problem in IMA. Begin by evaluating traffic on the network and the XML Broker.
|Name Resolution Web Server Duration (NRWD)||When this metric is high, there could be an issue with the Web Interface server or the XenApp plugin site (formerly known as the Neighborhood Agent site), the XML Service, the network link between the two, or a problem in IMA.
Like NRCD, this metric indicates how long it takes the XML service to resolve the name of a published application to a XenApp IP address. However, this metric is collected when a Web Interface site is performing this process on behalf of a launch request it has received from either the XenApp plugin (previously known as Program Neighborhood Agent) or from a user clicking a Web Interface page icon. This metric applies to all sessions launched through the Web Interface or the XenApp plugin (formerly, the Program Neighborhood Agent).
|Session Look-up Client Duration (SLCD)||This metric represents the time it takes to query every session to host the requested published application. The check is performed on the client to determine whether an existing session can handle the application launch request. The method used depends on whether the session is new or shared.|
|Ticket Response Web Server Duration (TRWD)||This metric is collected when Program Neighborhood Agent or Web Interface is the application launch mechanism. This metric indicates the time it takes to get a ticket (if required) from the STA server or XML service.||
When this metric is high, it can indicate that the Secure Ticket Authority (STA) server or the XML Broker are overloaded.
|Session Creation Client Duration (SCCD)||This metric represents the time it takes to create a new session, from the moment wfica32.exe is launched to when the connection is established.|
You can use this report in any of the following situations:
- After a user complains about how long it takes for his or her session to load
- Before rolling out an initial farm deployment
- After upgrading your farm or making significant changes
- Run the Suggested Startup Duration Detail report on the entire farm or a specific group of application servers, depending on what you are trying to troubleshoot.
- If you are trying to troubleshoot startup times for a specific user, specify his or her user name in the Optional Parameters.
By default, as illustrated in the screen capture above, this report groups information by device and user name. In the contracted view, the report displays the total average amount of time it takes in milliseconds for the session to start – see the Average Server Startup Duration (Avg SSD) metric time. Note that this is an average of all sessions listed beneath that user device when that line in the report is expanded.
- Expand the device so details for the device are visible.
- Using the screen capture at the top of this article as an example, reading from left to right you can see the values for the following:
| Column Name | Metric |
CASD Credentials Authentication Server Duration. The time (ms) spent on the server authenticating the user credentials. COSD Credentials Obtention Server Duration. The time (ms) taken for the server to obtain the user credentials. This is only likely to be a significant amount of time if manual login is being used and the server-side credentials dialog is displayed (or if a legal notice is displayed before login commences). PLSD Profile Load Server Duration. The time (ms) spent on the server loading the users’ profile. DMSD Drive Mapping Server Duration. The time (ms) spent on the server mapping the users’ client drives, devices, and ports. LSESD Login Script Execution Server Duration. The time (ms) spent on the server running the users’ login script(s). SCSD Session Creation Server Duration. The time (ms) spent on the server creating the session. The duration starts when the ICA client connection has been opened and ends when authentication begins. This should not be confused with the overall Session Startup Server duration.
In this report, for the first user, Jim S, it takes over 2 minutes for the session to start – as indicated by the Avg SSD metric in the far left column. The longest server-side tasks are for autocreating printers (PCSD), loading the profile (PLSD), and session creation (SCSD).
To try to reduce the session start time, the administrator’s best option would be to begin by investigating the reason it is taking so long for XenApp to create printers for the session. One place to start would be by finding out how many printers are mapped locally on the client device and calculating the resultant policy for that user when he or she connects from that device. If the user has a lot of local printers, consider editing the Printer Autocreation policy (in XenApp 5.0 and earlier) to limit the printers that are created.
After that, the administrator might want to investigate the profile configuration for this user. For more information, see User Profile Best Practices for XenApp 5_, CTX120285.
In this case, the amount of time dedicated to login scripts is negligible relative to the printer autocreation and the profile loading, so the administrator might choose not to investigate the scripts.
- Improving Slow Session Start Times. EdgeSight Wiki, accessed on October 1, 2009.
- User Profile Best Practices for XenApp 5, CTX120285. Citrix Systems.
- Improving visibility to the user login experience with Citrix EdgeSight. Citrix Systems, 2007.