Reduce the dataset for CMDB RELATIONSHIP HEALTH score computation

[Anonymous_guy]

Hi community,

Is there any way to exclude some classes while computing the relationship score? As far as I know, inclusion rules can be configured to exclude specific classes to populate normal health score that includes completeness, correctness and compliance KPIs.

Why do we need to achieve this?

   - A customer who has very limited CI classes in their business would never show interest in viewing relationship health score (duplicate, orphan and stale relationships) as a whole.

We are interested only in the 13k records shown above.

But the relationship dashboard has 100k records in it, that is nothing but the total number of records in the CMDB relationship table with UNDESIRED CIs (from a business perspective). Can we configure relationship score job to consider only a subset of this data?

Any help would be much appreciated.

[Sandeep90]

@Anonymous_guy We recommend to disable the relationship jobs fully as it has performance impact, from Xanadu it is disabled by default. We plan to remove this dashboard once we ship automated jobs to do the cleanup. 

"If either the CMDB Health Dashboard - Relationship Compliance Processor or CMDB Health Dashboard - Relationship Score Calculation dashboard job is active, that job is deactivated during the upgrade to Xanadu. After the upgrade is complete, you can reactivate those jobs to resume health reports for CI relationships. The active state of all other CMDB Health dashboard jobs is retained." 

https://www.servicenow.com/docs/bundle/xanadu-release-notes/page/release-notes/now-platform-capabili...

[Anonymous_guy]

@Sandeep90 Thanks for your information.
Can you please elaborate 'We plan to remove this dashboard once we ship automated jobs to do the cleanup"?
1) Does this mean dashboard will be removed forever from the platform? If yes, what would be the alternative to track the health?
2) What did you mean by cleanup? Deletion of unwanted relationships? 

We are concerned about the dataset for the Relationship score computation but the deactivation of relationship jobs makes no good. As far as the 'relationship score' tab is available, we are obligated to the client for the occurrence of a big numbers.

Thanks in advance

[Sandeep90]

Relationship health simply reflects the status of relationships based on the failure of health CIs, which may not always be appropriate. While you have the data, there are no direct action items you can take on the relationship table to resolve these issues. Instead, addressing the CI health failures will automatically resolve these concerns.

[Anonymous_guy]

@Sandeep90 Thanks for responding.

Let me make my question more clear. Let's suppose I'm working for a client who is only interested in computer and its extended classes. So, obviously, my client wouldn't be interested in viewing the health status of other classes. Hence, we would use inclusion rules to reduce the dataset for CMDB HEALTH SCORE - which is fine. But when it comes to the RELATIONSHIP HEALTH SCORE, the system considers the complete data residing in the relationship table (includes other classes as well) while computing the score. Do we have any means to include only the relations of computer and its extended classes just like how we use inclusion rules for CMDB HEALTH? 

I got your point that we need to look at the failures for improving the score. But how can we restrict other classes from score computation? For eg: If printer class has 100 duplicate relationships, why would we even consider those 100 for score computation since my client's interest is only on computer and its extended classes? Mitigation of duplicates is a second issue once we set the correct dataset.
To summarize, I'm not worried about the failures at the moment, but the total number of records considered.

Any suggestions?