Amsuingly the article compared to Pandas. Which in 2.0 has an Arrow backend. It'd be fun to see an update with the new Arrow backend. The idea of having a zero

Amsuingly the article compared to Pandas. Which in 2.0 has an Arrow backend. It'd be fun to see an update with the new Arrow backend. The idea of having a zero-copy datastore is very compelling. I'd futzed around a little with Arrow Ray, which is just a dumb shared-object store. Going further makes hea sense. It also suggests the need for radically higher radix db's, where most nodes have some local db instance, that selectively pulls in data. With the disaggregated db model that's rapidly rising in popularity, this comes semi-free.