How many of you actually use Machine Learning Powered mapping?

[NOW_seeker]

We are doing Service Mapping in our organization in preparation for CMDB go-live.

So far we have done Dynamic CI Groups and some manual mapping. The topic of ML (Machine Learning) based mapping came up and the Architect chose to defer it.

Just curious — How many of you have actually used ML powered mapping and has it been beneficial in any sense? Also, what did you use it for?

Please describe your use-case.

Thanks

[Matthew_13]

We’ve used ML-powered Service Mapping mainly to automatically identify CI relationships in complex applications.

Our main benefits were:

• Saves time vs. manual mapping

• Finds relationships that might be missed manually

• Improves over time as the model learns

Our Limitations:

• Suggestions still need review

• Less useful for simple and static services

@NOW_seeker - Please Give a Thumbs Up and Accepted Solution if you found Helpful!!

[NOW_seeker]

@Matthew_13 — Has it ever failed to identify a valid connection? 

[Matthew_13 · Accepted solution]

Yes — it can miss or incorrectly infer connections in certain scenarios.

We’ve seen ML-powered Service mapping struggle when traffic is intermittent event driven, or encrypted end to end especially if there isn’t enough consistent network or process data for the model to learn from. It can also miss dependencies that are logical rather than technical (Like, batch jobs, message queues or integrations that don’t maintain persistent connections.

That’s why we still treat ML-discovered relationships as recommendations not authoritative truth. They work pretty well for complex dynamic applicationsn but they still need validation particularly for business-critical services where accuracy matters.

@NOW_seeker - Please Give a Thumbs Up and Accepted Solution if you found Helpful!!

[Matthew_13]

Hi my friend 🙂

Just following up to see if good and to mark my post as Accepted Solution Kindly.

@NOW_seeker