Low-disruptive and timely dynamic software updating of smart grid components

Abstract

Components in the power grid require security, high availability and real-time communications for reliable operation. But these components are based on software that contains issues that need to be fixed. Timely installation of software updates allows securing vulnerable software quickly but conventionally disrupts availability and communications. Rolling updates on redundant systems prevent such disruptions but delay update installations as they need to be prepared carefully to update reliably. Dynamic Software Updating shortens the installation duration of updates by implementing them in-memory, allowing timely hot-fixing and installation of new features without service disruption or degradation in soft real-time communications. As the Smart Grid settles on standardization and common technologies for interoperability, the need for timely hot-fixing and updating of software applications and libraries which are in widespread use increases. In this paper, we discuss requirements of Smart Grid components and their updating opportunities. Afterwards, we present Lusagent, our dynamic updating system for Java 6 to 8 that is based on a novel eager program state transformation approach. We illustrate its programming efforts in a case study on an open-source Java control system framework and on several other server applications. Furthermore, we present performance measurements of dynamically updating these applications. The results demonstrate the potential of our dynamic updating approach in enabling low-disruptive and timely updating of highly available and widespread components at low and only one-time programming efforts.