What is the best process practice to perform the following requirements?
Please advise me on this.
Custom properties clean up is incredibly vague. Do you intend to delete properties that are no longer in use, update properties that are incorrect, create new properties to more effectively organize what you have? Could take an afternoon or a month depending on the goals and the quality of data you have available to identify mistakes and supply corrected data.
Retention, is that 10 million rows across all historical tables or 10 million in each? Historical tables is a function of object count, polling interval, and retention period so for example if you had 10k network interfaces in your environment you have 10k data points added every 2 min, then you'd hit that threshold in less than 2 days on the interface availability table. If you have a small environment then that might give you months of detailed retention.
Cleanup, I find that this is a process issue, how are nodes and interfaces being added to the monitoring. How do things get removed? What do you think needs to be cleaned up, how do you know when a node has been decommissioned?
Answering those questions can be easy with a good change control process and cmdb, or they can be nearly impossible in a loose, ad hoc environment with employees out just winging things at separate locations.
