Realizing software longevity over a system's lifetime

A successful instrument or telescope will measure its productive lifetime in decades; over that period, the technology behind the control hardware and software will evolve, and be replaced on a per-component basis. These new components must successfully integrate with the old, and the difficulty of that integration depends strongly on the design decisions made over the course of the facility's history. The same decisions impact the ultimate success of each upgrade, as measured in terms of observing efficiency and maintenance cost. We offer a case study of these critical design decisions, analyzing the layers of software deployed for instruments under the care of UCO/Lick Observatory, including recent upgrades to the Low Resolution Imaging Spectrometer (LRIS) at Keck Observatory in Hawaii, as well as the Kast spectrograph, Lick Adaptive Optics system, and Hamilton spectrograph, all at Lick Observatory's Shane 3-meter Telescope at Mt. Hamilton. These issues play directly into design considerations for the software intended for use at the next generation of telescopes, such as the Thirty Meter Telescope. We conduct our analysis with the future of observational astronomy infrastructure firmly in mind.