adatole wrote:
To my knowledge, the device stats (CPU, RAM, etc) are purged the night after the delete. Meaning:
When you delete the node, the records from the nodes table (and possibly the "holy trinity" tables - nodes, volumes, interfaces - are moved to a temporary "deleted" table (relatively easy to find if you dig using SQL Management studio or similar toolset).
Then during the nightly maintenance routines, there is a scan for orphan records in each table, and the stats are really and truly deleted.
To circumvent this, we have a routine that pulls the daily detailed stats out to a separate SQL database that has massive storage. Yes, it's a big bad boy. But we only use it for long term data retention, reporting, forensics, etc. The point was to be able to keep Solarwinds nimble but also have the details you are talking about.
Wow that's keen! I'd just say no to anyone wanting old info that wasn't obtainable easily enough 
. However, I do unmanage devices for about 1 month when they are decommissioned (not really the same thing as what OP has done) just in case they are put back into production.
But when I replace a device, I will usually just update its details and let it rediscover itself so old data is still available.