What type of node should I run for specific use cases?

I want to run a node. I would like to be able to pull historical state from smart contracts, in particular a bunch of liquidity pools. I'd like to run my own node that I can query for this information instead of using infura or something like that. I get all kinds of conflicting information when I search for information on this, including a ton of outdated information. I'm not sure if I should run an archival node, or what level of pruning I should use, what the levels even are, and if theres a way to prune but keep historical data for particular contracts, if I can sync data from other nodes as needed.

I also might be interested in running nodes for L2s and rollups and what not depending on the upside.

Does anyone here have detailed information on this or know where I can find it?