@eileennoonan for reference, we've got a somewhat complex schema and some inefficiencies in our infra, but we're seeing p50 in prod for LR of between 20 and 300ms