I'd just write a bunch of adapters to transform all your current data into a unified set of formats in acro or parquet or whatever, compress the originals and d

I'd just write a bunch of adapters to transform all your current data into a unified set of formats in acro or parquet or whatever, compress the originals and dump into glacier just incase and then use Athena to query it. Then you can have s3 file write triggers to ensure that every new file conforms to your new schemas