I am sure that I am not the only one who often bemoans that their university doesn't have that [extremely specific book or paper] they're looking for.
On reflection, I realise it is impractical for the library to have a copy of every mathematical text or paper...but just how impractical would it be?
To a rough approximation, how many textbooks and papers have been written on modern higher mathematics? How many new ones are published each year? How large a library would be needed to hold them all?