If we can cache one PrepareStatement's corresponding execution plan like
the traditional relational database, we can save the logic and physical
execution plan time. I want to know what's the potential difficulties . If
we list these difficulties out ,we may find some solutions to solve it
directly or indirectly .

I give one first:
1.  to cache physical plan :
   the corresponding underlying data distribution changed  leads to a
different join or aggregate strategy

Reply via email to