To put is cynically, everyone knows that yesterday's tools don't work. So it's better to try today's tools, which you don't yet know don't work. This allows one

To put is cynically, everyone knows that yesterday's tools don't work. So it's better to try today's tools, which you don't yet know don't work. This allows one poorly thought out idea to replace another - especially when selling new stuff is profitable. That said, it's implausible that the main problem is scaling since mainframes scaled considerable data quite a while back.