Performance properties of time-series operations in R at scale (mainly xts and data.table)

Question

I am undertaking a new project encompassing large time-series datasets from which dependent calculations are fed into a shiny application. As such, efficiency i