How do you manage data sources for Normalization (all in one process or extra processes)

[dennis_reinhardt]

Hi @ll,

In the admincConsole you have the possibility to create several processes for the inventory of data sources. Or you just create one process and add all your data sources into it.

Unfortunately I didn't find any information in the help about when to create a separate process for each source and when to combine multiple sources in one process.

All sources in a process have the "advantage" that the graphic of the normalization looks nicer 🙂 Currently I have one process each for SCCM and FNMS.

Any advantages or disadvantages for these two options?!

Thanks and Best,

Dennis

[MurrayPeters]

Hi Dennis, 

One of the main reasons for having multiple sources in one process is because you want to de-dupe the records - for example if computer with serial number XYZ0001 exists in both FNMS and SCCM then Normalize will identify that accordingly.   

If however you only ever want to look at SCCM data in isolation, then you could use a separate process.

-Murray

[dennis_reinhardt]

Hi  @MurrayPeters 

Thanks for your feedback.

If I understand your posting correctly, then a duplicate check only takes place within a process. If you use multiple processes and data sources, do the devices appear twice in BNDA?

Example:
Process A (SCCM & FNMS)
- SCCM Device with SN 1234
- FNMS Device with SN 1234
Result: 1 Device wih SN 1234 & all Information from FNMS and SCCM

Process B1 (SCCM)
- SCCM Device with SN 1234
Process B2 (FNMS)
- FNMS Device with SN 1234
Result: 2 Devices wih SN 1234

Thanks an Bst,

Denni

[Jason_Lilleboe]

@dennis_reinhardt  

You are correct.  I have seen many customers uses processes to separate out data from Production and Test environments.   Additionally, you can use the INV Name property as a filter in the User Console to report separately on the two processes.

As for the rules around the deduplication rules see:  https://community.flexera.com/t5/Data-Platform-Knowledge-Base/Deduping-rules-in-Normalize/ta-p/535/jump-to/first-unread-message.

- Jason

[dennis_reinhardt]

Hi @Jason_Lilleboe ,

Thanks for your replay and the detailed answer. Will take a look into the INV NAME topic. When I later export the data into a 3rd party system (FNMS, CMDB, ServiceNow) - is it possible to decide which "process data" I would like to export. As I understand, when connecting BDNA to a 3rd party system, the import will take the whole BDNA_PUBLISH DB.

Question regarding to your link ...

Does "To eliminate duplicate records from your inventory ..." means, all data linked to this device (HW & SW information) from the other inventory source isn't processed/used by BDNA - or is just to avoid multiple devices with the same Name/SN, but includingt he full set of information over all sources.

Thanks and Best, Dennis

[JohnSorensenDK]

Hi @dennis_reinhardt,
The pre-built integration configurations for import into 3rd party systems will typically not include any filtering based on inventory source, i.e. you'll have to change the configurations according to your needs. (They don't necessarily include the entire BDNA_PUBLISH database though depending on the integrations...)

De-duplication is built in to ensure that devices imported from multiple inventory sources get de-duplicated but still include all information from all inventory sources. The prioritization of the inventory sources controls which inventory sources overrules others when discrepancies are found:

Thanks,

[dennis_reinhardt]

Hi @JohnSorensenDK ,

as I've learned from your collegague @ajeffries, at the moment DP doesn't "include all information from all inventory sources." it will ignore data from other sources when the device is already imported from a higher prio data source. Could you please check this statement and send me an update.

Thanks and Best,

Dennis

[JohnSorensenDK]

Hi @dennis_reinhardt,

I was describing how it works in general, but if you have discussed specific needs I would recommend continuing these discussions directly. (There are other options like pass-through if you need something specific from a data source...)

Thanks,