Application Usage - only reporting 30% usage on commerical applications
Hoping you can help.
We are currently reviewing the usage for commercial applications within our environment. At present only 30% of commercial installation applications are reporting usage information. The other 70% of installed commercial applications are reporting "0" usage.
I need guidance on how to understand if the 70% not reporting usage can be improved and what checks can be carried out to ensure usage reporting is possible for that application.
- Flexnet Invenroty agent version 14.2 is installed
- The settings for UseAddRemove - has been set to true on the agent to report usage
1) How do we check the appropriate installer of file evidence is linked to the application that match usage metering data?
2) How do we identify which applications have limitations regarding usage metering - and report '0' usage cannot be improved.
3) We have assumed - if an application is reporting within the installed Application menu; the evidence is linked to the application as installation is reporting - and therefor usage evidence will also be linked - that the installer/file evidence is not separate for usage..
a) If not correct with this assumption, please detail how do you check recognition rules for usage? If the application is reporting installed..
4) Any reports that we can be run for all applications to show more in-depth information about what applications are reporting usage and can report usage?
5) Checking each of the applications within that 70% no usage reporting list individually is not an answer.. or is it?
Many thanks in advance,
Usage tracking is not so easy, quick answer, Flexera will complete:
- best is to use last version of agent and last version FNMS (various bugs in former versions).
- usage tracking is based on file evidence (binaries), not install evidence.
- link between file evidences and application can be seen in application / file evidence display.
- link between file evidence and product can be seen in raw usage report
- be aware, raw usage report is a weekly report, you have to consider weeks, days are not always accurate.
The data flows and analysis involved in usage can be fairly complex. The "30%" figure noted in this post doesn't particularly surprise me, but there isn't a single simple answer to magically get more usage data. Take a look at the following post which describes many ins and outs related to usage metering capabilities when using the FlexNet inventory agent: Application usage metering using the FlexNet inventory agent.
In relation to checking appropriate evidence is linked to application records: the Discovery & Inventory > Raw Software Usage page is a key place to start when using the FlexNet inventory agent to meter usage. Search on this page for raw details related the application in question - you will see file and possibly installer name details here. You can then look up to see whether file and installer evidence rules are configured in the ARL to recognize those details.
In relation to installer vs file evidence: The recognition process can recognize usage based on both installer evidence and file evidence. Whether installer evidence or file evidence data is available in the source system that metered the usage in the first place will depend on the capabilities of the tool you are using. If the FlexNet inventory agent is used for metering usage then:
- All usage is imported as file evidence.
- Most usage is also imported as installer evidence. An exception is that latest agent versions do not report installer evidence for a used executable that cannot be unambiguously associated with specific installer evidence (such as when the executable is installed under a directory where multiple applications are installed).
Some things to consider to identify whether usage gathering for a particular application can be improved through further configuration are:
- Check in the Discovery & Inventory > Raw Software Usage page to identify whether the agent is actually metering usage for executables known to be associated with the application. Some common reasons the agent will not gather usage for an application are:
- The executable does not meet the conditions required to be tracked. That is, it is not under a directory identified as being associated with a package installed for all users, or identified as the key file in a Windows Installer package component, or identified to be tracked using Manual Mapper configuration settings.
- The executable is not run as a user-process (for example, it may be a system or non-interactive process).
- The application isn't directly invoked by running a .exe file (for example, it is a plugin loaded by some other process).
- If the agent is metering usage, check whether a file and/or installer evidence rule is configured that will recognize the evidence shown on the Raw Software Usage page.
Thanks @ChrisG appreciate your response.
Check in the Discovery & Inventory > Raw Software Usage is definitely a slow process.
Tackling all commercial applications via raw software usage will be time consuming process.. any tips on descoping apps that will not report usage as a starter to reduce the list of applications to start investigation on Evidence.
Or can Flexera provide a list/publishers known not to report usage?
You could use the number of installations of an application with no usage reported as an heuristic to indicate the likelihood that usage won't be tracked (with Manual Mapper configuration) for that application: if an application is widely installed then it is likely somebody is using it, so no usage being detected is a reasonable indication that usage won't be tracked for it.
That's only an heuristic though and there will be exceptions: it is possible some software is widely deployed but really is not used.
I can't think of any way to distinguish with precision en masse between installed applications that (a) have not been used, and (b) will not return usage (without Manual Mapper configuration).
One other consideration is that it is also possible that not all usage of an application is tracked even though some usage of the application is detected. A couple of examples I've recently come across are:
- Actual Microsoft Teams usage may be reported on some computers but not other computers. This is due to the way its installer works: usage metering will track execution of the Teams installer process, but may not track the interactive Teams process itself.
- Some usage of Foxit PhantomPDF executables may be tracked while usage of other executables related to this application may not be tracked. This is due to the way the PhantomPDF installer installs different .exe files. Some files are installed in a way that meets the conditions required to be tracked, while other files are installed in a way that doesn't meet the conditions to be tracked: they are not Windows Installer component key files, so are not identified as being part of the PhantomPDF application.
The upshot of all of this is that it is important to treat usage data as indicative of what is going on, and not an exact or complete representation. All else being equal, it is safer and more appropriate to make a statement like:
"We have found application XYZ is installed on your computer but our system has not detected it being used - does that match your understanding?"
rather than an absolute statement like:
"We have found application XYZ is installed on your computer but it has not been used".