Best practise to remove millions of stale CIs in CMDB

[Shilpa29]

Hi team,

We’ve enabled Discovery for Cloud, Container, Storage, and Network components. Upon reviewing recent data, I noticed that out of approximately 7 million records, around 4 million have not been discovered in the past 90 days. 

Below classes are the major contributors for this count. Typically, we retire CIs to preserve historical ticket data. However, the classes in question are not designated as principal classes, and I don’t foresee any issues with deleting these CIs and their relationships. I am just wondering if anyone have done similar clean-up before ? and are there any best practices to follow? is anything that I need to be aware before deleting them? mainly the hardware types and IP address classes?

Class Name	CI Count	Owner
Cloud Resource	3120737	Cloud
Hardware Type	214714
OpenShift Image	182009	Containers
Kubernetes Volume	75684	Container
OpenShift Docker Image Repository	62821	Containers
Storage Device	41043	Storage
IP Address	38289	General
Disk	36020	Storage
Disk Partition	30968	Storage

[Swapna Abburi]

Hi @Shilpa29 

You can use 'Delete jobs' under System Data Management or table cleaner (sys_auto_flush) feature to delete large volume. of records. it will take time for system to delete these records and these jobs run in background.

I think you can safely delete records in the listed CI classes based on last discovery date as we don't create incidents/CRs for these CI classes usually. Also, as far as I know, they don't contribute to CI relationships.

[Shilpa29]

Hi Swapna,

Thanks for your response. Appreciate it.

If I wanted to set-up a regular review and deletion, are you able to advise if the above approach works? like initiate deletion everyday if the CIs are not discovered for more than 90 days? I heard about Data management policies too, which one is the best approach for everyday review and deletion. 

[Swapna Abburi · Accepted solution]

@Shilpa29 

First time, when you are deleting them in bulk, you may go with delete jobs in Data management. Later, If you want to execute deletion as a schedule, then you can have a table cleaner (sys_auto_flush). Table cleaner job runs hourly and if any records found with the matching condition, it deletes them.

[Shilpa29]

Thank you so much Swapna