Software Aging and Rejuvenation: Where we are and where we are going Domenico Cotroneo, Roberto Natella, Roberto Pietrantuono, Stefano Russo Dipartimento di Informatica e Sistemistica Universit` a degli Studi di Napoli Federico II Via Claudio 21, 80125, Naples, Italy {cotroneo, roberto.natella, roberto.pietrantuono, sterusso}@unina.it Abstract—After 16 years, a significant body of knowledge has been established in the area of Software Aging and Rejuvena- tion (SAR). In this paper, we survey papers about SAR that appeared in IEEE conferences and journals, identify where SAR research has been mostly focused, and highlight some aspects deserving more attention, with the aim to provoke a con- structive discussion among SAR researches about where SAR has arrived and where it should be headed in the next future. Keywords-Software Aging; Software Rejuvenation; Survey I. I NTRODUCTION Since 1995, year of publication of the seminal work of Huang et al. [1], much efforts have been devoted to characterize and mitigate the Software Aging phenomenon, that is, the accumulation of errors occurring in long-running operational software systems that leads to progressive resource depletion, performance degradation, and eventually to the hang or crash of the software system. As a result, a significant body of knowledge has been established and an international community of researchers in the area of Software Aging and Rejuvenation (SAR) has grown. Therefore, after 16 years, it is reasonable to look at what has been made, what has still to be accomplished to transfer the results to the industry world, and which are the future challenges for the SAR community. Starting from the analysis of 71 papers, which appeared in IEEE conferences and journals, this paper tries to identify where SAR research has been mostly focused, and to highlight some aspects which still deserve more attention by the SAR community. The aim is to stimulate a constructive discussion among SAR researches about where SAR has arrived and where it should be headed in the next future. In Section II, we describe how SAR research papers have been collected. Section III reviews SAR literature with respect to four dimensions. Section IV concludes the paper with open issues and research opportunities. II. ANALYSIS OF LITERATURE To have a picture of the current status of SAR literature, we analyzed some of the most important conference proceedings and journals. The steps followed for the analysis are: Search engine. We relied on IEEE Xplore (http://ieeexplore.ieee.org/) to conduct the analysis. Selection of conferences and journals. This preliminary analysis focused on the most relevant conferences/journals in the field of dependability, also considering the workshops held jointly with them: International Conference on Dependable Systems and Networks (DSN), International Symposium on Fault Tolerant Computing (FTCS), Pacific Rim International Symposium on Dependable Computing (PRDC), International Symposium on Software Reliability Engineering (ISSRE), International Symposium on Reliable Distributed Systems (SRDS), Transactions on Software Engineering (TSE), Transactions on Reliability (TR), Transactions on Computers (TC), Transactions on Dependable and Secure Computing (TDSC). In addition, we also considered part of the software engineering community (by querying for “software engineering” in the proceedings or journal title), addressing conferences/journals like International Symposium on Empirical Software Engineering (ISESE), International Conference on Software Engineering (ICSE), World Congress on software Engineering (WCSE), and other minor conferences. Search criteria. The research has been carried out by querying for the words “aging”, or “rejuvenation”, or “leak” in the metadata of the IEEEXplore research engine (e.g., title, abstract, keywords, etc.). Note that in the case of software engineering, the word “aging”, as well as “rejuvenation” has a different meaning: it indicates the software becoming obsolete, e.g., because of changed requirements, maintenance actions, and so on. As a result, most of the papers found in that case is not related to software aging as intended by the dependability community. Actually, only 8 papers of software engineering community are related to SAR. It should be noted that some papers on SAR appeared in venues not published by the IEEE or related to other computer science fields, and that we focus our analysis on the ones previously mentioned. In Figure 1, the number of papers per year is reported. As may be noted, since 1995, the trend is increasing. A sharp increase of scientific productivity of the SAR community was fostered by the 2008 WoSAR workshop, which took place for the first time in that year. The venues preferred for SAR studies were ISSRE (30), DSN (12), and PRDC (8). III. WHERE WE ARE Past work on SAR can be analyzed with respect to several dimensions. We here consider four dimensions, namely: • the type of analysis that has been conducted, • the type of system to which the work is related, 2011 Third International Workshop on Software Aging and Rejuvenation 978-0-7695-4616-2/11 $26.00 © 2011 IEEE DOI 10.1109/WoSAR.2011.15 1