Keynote Speakers

Keynote 1 

Software Aging and Rejuvenation Through the Decades: Evolving Challenges and Emerging Strategies in the Serverless Era  

Rivalino Matias

Abstract 

Software systems have undergone a remarkable architectural evolution—beginning with tightly coupled monolithic applications, advancing through multi-tier client–server models, and transitioning into today’s highly distributed and loosely coupled service-oriented ecosystems. Yet, despite advances in software engineering, software aging remains a persistent and inevitable phenomenon, silently degrading system performance, reliability, and availability over time.

This keynote will trace the historical trajectory of Software Aging and Rejuvenation (SAR), examining how aging manifests differently across architectural paradigms and how rejuvenation strategies have adapted to each new wave of computing. We will discuss lessons learned from decades of SAR research and practice, emphasizing what remains fundamentally unchanged and what must evolve.

As we move into an era dominated by serverless computing, traditional SAR assumptions are being challenged. What does aging look like in such ephemeral computing environments? Can classical rejuvenation techniques still apply, or do we need new ones?

By bridging past wisdom with emerging trends, this talk will provide both researchers and industry practitioners with a comprehensive view on how to understand and design aging-aware systems in modern computing environments.

Biography 

Dr. Rivalino Matias is a professor and researcher in Computer Science. He holds a PhD and is a tenured Associate Professor at the School of Computer Science, Federal University of Uberlândia (UFU), Brazil.

His research primarily focuses on Software Reliability, with the main projects delving into the "Software Aging and Rejuvenation" and "Reliability of Operating Systems." He has an extensive publication record, with numerous peer-reviewed papers, book chapters, and best-paper awards. As an expert in his field, he has contributed to the understanding and advancement of software dependability. 

Dr. Matias brings a unique blend of academic and industry experience, having worked in the IT industry for over 20 years for companies such as Microsoft, Red Hat, and IBM, across Brazil, Latin America, and the USA. He also provided specialized consulting services to various sectors of the Brazilian federal government, the European Union, and private initiatives.

Keynote 2
Modeling Time-to-Failure Distributions Across the Bathtub Spectrum Using Phase-Type Models   

Paulo Maciel

Abstract 

Aging systems often exhibit time-to-failure behaviors that vary over time, including decreasing failure rates during early life (infant mortality), constant rates during their useful life, and increasing rates as components wear out — collectively forming the well-known bathtub curve. Modeling these behaviors accurately, while maintaining tractable models, is a central challenge in reliability and availability evaluation. This keynote presents a data-driven approach for modeling such systems using Phase-Type (PH) distributions. The strategy maps decreasing failure rates to hyperexponential or Cox-2 distributions, constant failure rates to exponential distributions, and increasing failure rates to Erlang, hypoexponential, or Cox-1 distributions. To capture the entire bathtub profile, Generalized Erlang and general Coxian distributions may also be employed due to their flexibility and expressive capacity. Two distinct parameter estimation approaches are considered: moment matching, which provides a fast, closed-form solution based on empirical moments; and the Expectation-Maximization (EM) algorithm, which is particularly suited for accurately fitting Generalized Erlang distributions. These methods not only yield the underlying parameters but also provide full analytical characterizations, including the probability density function (PDF), cumulative distribution function (CDF), and hazard function. Additionally, they enable the automatic generation of equivalent Continuous-Time Markov Chains (CTMCs) or Stochastic Petri Nets (SPNs) that represent the fitted distributions, facilitating formal modeling and analysis. The talk includes illustrative figures and examples from the two-volume reference Performance, Reliability, and Availability Evaluation of Computational Systems, which help clarify how the described techniques can be applied in practice. These methods are discussed in the context of typical modeling challenges, with applications related to redundancy planning, software rejuvenation, and dependability assessment over time.
 
Biography
Paulo Maciel graduated in Electronic Engineering in 1987 from the University of Pernambuco and holds a master's and a doctorate in Electronic Engineering and Computer Science from the Federal University of Pernambuco. During his doctorate, he completed a "sandwich internship" at Eberhard-Karls-Universität Tübingen, Germany, from 1996 to 1997.

Paulo Maciel was a professor at the Department of Electrical Engineering at the University of Pernambuco from 1989 to 2003. Since 2001, he has been at the Computer Center, Federal University of Pernambuco, where he holds a full professorship. He also serves as a member of the National Council for Scientific and Technological Development (CNPq - Researcher productivity grant category: 1A).

In 2011, he took a sabbatical year at the Department of Electrical and Computer Engineering, Edmund T. Pratt School of Engineering, Duke University, USA.

Paulo's research interests include performance, reliability, availability, capacity planning, and stochastic models, with applications in cloud computing, sustainable data centers, manufacturing, integration, and communication systems.

He has worked as a consultant and research coordinator on projects funded by companies such as the US Army Research Office (ARO), HP, EMC, CELESTICA, FOXCONN, ITAUTEC, INdT, Samsung, Motorola, CHESF, and Brazilian government research agencies including FACEPE, CNPq, and FINEP. Currently, Paulo coordinates projects funded by FACEPE, CNPq, and FINEP.

In 2023, he published the two-volume book Performance, Reliability, and Availability Evaluation of Computational Systems (CRC Press), which provides a comprehensive treatment of modeling, measurement, and data-driven analysis techniques in the area of dependable computing.