People of ACM - Luciano Baresi

October 22, 2024

You have worked on a wide variety of topics within software engineering. Is there a common thread that runs through the kinds of challenges you take on?

There are different reasons for these different topics. The discipline has evolved over the years, together with new challenges and research directions. For example, service-oriented computing did not exist when I started my career. I am a very curious person, and I am always interested in new problems from industry, new topics, and new challenges. I have often decided to change my interests to avoid doing the same things forever and use the new topics as a means to keep myself up-to-date. This approach helps me to publish good papers, attract new collaborators, and be competitive for new grants.

Along with co-authors, you published the paper, “NEPTUNE: network- and GPU-aware management of serverless functions at the edge,” which won the Best Paper at the SEAMS conference in 2022. The paper pointed out that a wide range of software applications are constrained by cloud infrastructure. Will you briefly explain the technical challenge of cloud infrastructure and the insight your paper outlined to address it?

The paper speculates on the differences between cloud and edge computing. Cloud infrastructures are great in many cases, they help solve many problems related to the computing infrastructure, but they cannot address pressing latency constraints. Edge computing allows one to execute code closer to the user, and thus can address latency issues, but requires more sophisticated management capabilities because of pressing limitations and very dynamic workloads. The paper proposes a first solution to address these issues where edge nodes offer both CPUs and GPUs to facilitate the execution of AI-based computations, and our edge framework Neptune, manages them dynamically and proactively.

One of the most cited recent papers in the ACM DL you co-authored is “A Unified Model for the Mobile-Edge-Cloud Continuum.” Generally speaking, what kinds of advances will be needed for the software engineering community to keep up with the growth of mobile computing?

Users are not interested in knowing where their applications are executed, but they want to have responsive, efficient, and privacy-aware solutions. Since our mobile devices cannot execute everything (for very different reasons), and the cloud is not always able to accommodate every user requirement, a computing infrastructure that manages the computations transparently is mandatory. This means that the user can enjoy each computing platform (cloud, edge or mobile devices) without taking care of the details. The switches between them are managed automatically and properly to keep the correctness of ongoing computations and offer the best user experience.

Why was a publication such as ACM PACMSE needed? Why should researchers submit their work to be included in PACMSE?

PACMSE is a new opportunity for the community. It does not aim to compete with the existing top software engineering journals, but it offers a new means to publish novel and top-level research timely in a high-quality journal. Many conferences have evolved their selection processes towards what journals have been offering for years, and thus we think that PACMSE can be a better, more appropriate framework to continue this transformation. The pandemic and many other issues have also exacerbated the problem of decoupling publication and presentation at major international venues. PACMSE is also a Gold Open Access journal, and we think this is another great service to the community.