Distributed control systems manage complex production processes in chemical plants, oil refineries, and power plants. These systems span embedded, server-side, desktop, mobile, and cloud software and may consist of more than 10 million lines of source code. They typically execute for 15-40 years to support long-running production processes and are therefore continuously challenged by software aging and feature evolution. This talk provides an overview of different approaches towards sustainable software systems, which have been applied at ABB Corporate Research in the last few years. They range from quantitative software architecture measurement, over performance prediction for changing workloads, to architecture decision modeling. While software evolution remains a constant challenge, these approaches may help practitioners and researchers to mitigate issues of architectural technical debt and thus save software maintenance costs.