I would love to have the ability to define "column groups" in regular Postgres. For example most of the time I just leave the table row-oriented, but when tryin

I would love to have the ability to define "column groups" in regular Postgres. For example most of the time I just leave the table row-oriented, but when trying to optimize performance of specific tables and queries I could tell postgres to store one or a handful of columns separately similar to a column oriented database. This would allow getting quite a bit of performance out of my database for the few cases in OLTP where I need to do some scans or searches. Even just breaking out some rarely used columns into a separate group could be quite nice for keeping the caches hot.