Staffing Strategies for Maintenance of Critical Software Systems at the Jet Propulsion Laboratory

Abstract

Context: The Mission Design and Navigation Software Group at the Jet Propulsion Laboratory (JPL) maintains mission critical software systems. We have good empirical data and models for maintenance demand---when defects will occur, how many and how severe they will be, and how much effort is needed to address them. However, determining the level of staffing needed to address maintenance issues is an ongoing challenge and is often done ad-hoc. There are two common strategies are (1) reactive - add/remove staff as needed to respond to maintenance issues, and (2) capacitive - retain a given staff size to address issues as they occur, proactively address issues and prevent defects. Goal: To use our empirical models for maintenance demand to address the issue of staffing from a risk perspective. Method: We develop a staffing model that allows us to simulate large numbers of maintenance histories. From these histories we examine the risks posed to the institution as a function of the staffing available to address issues as they arise. Results: We find that the model developed matches our intuition. There is a "sweet spot" in staffing levels that allows issues to be addressed in a timely fashion. Below that level the institution experiences substantial risk; staffing above that level does little to improve the institutions risk exposure. Conclusion: The models developed provide tools that, for the first time, allow us to quantitatively discuss the level of staffing needed to ensure that we can meet the time constrained demands for maintenance on mission critical systems and thereby determine staffing budgets that ensure mission success.