The concept of AIOps has emerged to reduce the cognitive load and improve the productivity in the exploitation of large and diverse operational data produced during DevOps activities. This involves the use of data-driven models to expedite and automate the resolution of intricate IT-Envs problems. While the concept of a digital twin of a physical system has been around for a long time, the notion of a digital twin of a digital or software system has only emerged recently. This project investigates the use of digital twins (DTs) to explore AIOps scenarios proactively. Our goal is to combine AI, simulation, and experimentation-at-runtime techniques to develop a DT framework that helps development teams to observe, measure, model, and study the past, present, and future behaviour of RT IT-Envs.

Public Impact Statement:
The rise of advanced IT environments (IT-Envs) that meet ever increasing user expectations on software quality demands innovative practices in the development and operation (DevOps) of software-intensive systems [1]. DevOps teams seek to deliver value by addressing operational challenges that tend to overwhelm human capabilities. Most of these challenges relate to the structural and behavioural complexities of modern IT-Envs [1]. While the former concerns the orchestration of multiple technologies, the latter involves the exploitation of large data streams produced that are crucial to DevOps activities. As automation, autonomy, and artificial intelligence technologies are maturing and permeating various activities in the software development lifecycle, opportunities arise from their integration with DevOps practices to improve risk mitigation, root cause analysis, problem resolution, and operational optimization in IT-Envs [2].
The concept of AIOps has emerged in industry to reduce the cognitive load and improve the productivity in the exploitation of large and diverse operational data produced during DevOps activities. Recent progress in machine learning, automation engineering, and reliability engineering have given rise to many opportunities for reasoning and proactive AIOps [3] [5]. This involves the use of data-driven models to expedite and automate the resolution of intricate IT-Envs problems, thereby reducing the management complexity for human operators [1]. While the concept of a digital twin (DT) of a physical system has been around for a long time [6], the notion of a DT of a digital or software system has emerged only recently [2]. In this case, DTs can be leveraged to explore AIOps scenarios proactively. Thus, a DT can mirror relevant structural and behavioural characteristics of, for example, computing infrastructure---a real twin (RT) of a software system. By combining AI, simulation, and experimentation-at-runtime techniques, DTs enable development teams to observe, measure, model, and study the past, present, and future behaviours of RTs in IT-Envs [2].
This project focuses on the exploration and development of DT technology enabling proactive AIOPs. We concentrate on the creation of DT-driven simulation environments that enable replicating DevOps activities and systems to synthesize data and train AI models that contribute to mitigating the risks associated with continuous software releases in IT-Envs.
References
[1] IBM Cloud Education. 2020. AIOps. https://www.ibm.com/cloud/learn/aiops. Accessed: 2021-09-15
[2] H. A. Müller, L. F. Rivera, M. Jiménez, N. M. Villegas, G. Tamura, R. Akkiraju, I. Watts, E. Erpenbach. 2021. Proactive AIOps through Digital Twins. CASCON x EVOKE 2021. ACM. https://dl.acm.org/doi/abs/10.5555/3507788.3507833 Accessed: 2022-11-15.
[3] R. Akkiraju. 2021. The Art of Automation: Chapter 5 - AIOps. IBM. https://www.ibm.com/cloud/blog/art-of-automation-chapter-5 Accessed: 2022-11-15.
[4] S. Waterworth. 2018. Stan's Robot Shop - a Sample Microservice Application. Instana. https://www.instana.com/blog/stans-robot-shop-sample-microservice-application/ Accessed: 2022-11-15.
[5] Y. Dang, Q. Lin, and P. Huang. 2019. AIOps: Real-World Challenges and Research Innovations. In Proceedings of ICSE-Companion 2019. ACM: 4 -5. https://ieeexplore.ieee.org/document/8802836 Accessed: 2022-11-15.
[6] M. Grieves. 2015. Digital Twin: Manufacturing Excellence through Virtual Factory Replication. Technical Report. Dassault Systèmes. https://docplayer.net/37776975-Digital-twin-manufacturing-excellence-through-virtual-factory-replication.html
[7] J. W. Creswell and J. D. Crewell. 2018. Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. Sage Publications, 5th edition.
[8] J. Webster and R. T. Watson. 2002. Analyzing the past to prepare for the future: Writing a literature review. MIS Quarterly, 26(2):xiii -xxiii. https://www.jstor.org/stable/4132319 Accessed: 2022-11-15.
[9] I. Vessey, V. Ramesh, and R. L. Glass. 2005. A unified classification system for research in the computing disciplines. Information and Software Technology, 47(4):245-255. www.sciencedirect.com/science/article/abs/pii/S0950584904001260.
[10] S. Easterbrook, J. Singer, M. A. Storey, and D. Damian. 2008. Selecting Empirical Methods for Software Engineering Research, pp. 285 -311. In [12].
[11] A. Jedlitschka, M. Ciolkowski, and D. Pfahl. 2008. Reporting Experiments in Software Engineering, pages 201 -228. In [12].
[12] F. Shull, J. Singer, and D. I. K. Sjøberg, editors. 2008. Guide to Advanced Empirical Software Engineering. Springer-Verlag.

Research team: