I also write for the Communications of the ACM (CACM). I recently posted an essay with this title on the CACM blog. The essay, located here, discusses the challenges of sustaining cyberinfrastructure for scientific discovery.
As I noted in the essay, the longitudinal value of research data often accrues to disciplines other than those where it was created. We need a new model that judiciously balances investment between the "next big thing" and sustainable cyberinfrastructure, for the latter is itself a crucial enabler of discovery.