Q&A with Dimitrios Skarlatos, visiting scientist at Fb and assistant professor at Carnegie Mellon College
In this monthly series of interviews, we put members of the academic community and their critical research in the spotlight – as thought partners, collaborators, and independent contributors.
For October we have nominated Dimitrios Skarlatos, Assistant Professor in the Computer Science Department at Carnegie Mellon University (CMU). Before he took up his professorship at the CMU, Skarlatos decided to spend some time as a visiting researcher on Facebook. He has received several awards for his research, including a Facebook Faculty Award and the 2021 ACM SIGARCH & IEEE CS TCCA Outstanding Dissertation Award for “Contributions to the redesign of the abstractions and interfaces that connect hardware and operating systems”.
In these Q&A, Skarlatos shares his experience as a visiting scholar on Facebook, his motivation for a prebattical, the research projects he has worked on, advice to academics considering a similar path, and more.
Q: Tell us about your academic experience so far. What are your primary research interests?
Dimitrios Skarlatos: I’m an assistant professor in the Department of Computer Science at Carnegie Mellon University. My research combines computer architecture and operating systems with an emphasis on performance, security and scalability. My current work pursues two central topics: (a) the detection of security gaps and the establishment of defensive measures at the boundary between hardware and operating system and (b) redesign of abstractions and interfaces between the two layers in order to improve performance and scalability.
Q: What inspired you to spend some time on Facebook before starting your professorship?
DS: My motivation was to learn more about large-scale, real-world data center deployments and systems. I believe this learning experience is invaluable and will help me grow as a researcher. Facebook maintains a flat organizational structure that enables cross-organizational, interdisciplinary research on an unprecedented level. Even during my first engagement at Facebook, I had the opportunity to speak to very diverse teams and learn more about their research challenges, which are rooted in running massive production systems for almost every technical area that interested me, including data centers, kernels, global load sharing , Machine learning hardware, serverless computing, etc. I found this to be unique on Facebook and an incredible opportunity for visiting scholars like me. Access to the best and brightest minds, coupled with the ability to solve real-world problems on Facebook, enables seamless cutting-edge research, which I am particularly pleased about.
Q: What research projects did you work on during your time at Facebook? Which teams did you work with?
DS: My time on Facebook started working with the RAS team and Andy Newel. RAS is Facebook’s regional resource allocator that is continuously optimizing. RAS introduces a novel capacity abstraction called reservations. Based on this abstraction, a two-step approach to scaling the resource allocation for all data centers in a region is required, in which a solver with mixed integer programming continuously optimizes the allocations of servers to reservations off the critical path, and a conventional container allocator performs the placement of Containers on servers in a reservation in real time. RAS provides guaranteed capacity taking into account random and correlated failures, data center maintenance, heterogeneous hardware and compound workload constraints. Further information on RAS can be found in our SOSP 2021 paper “RAS: continuously optimized regional resource allocation for data centers”.
Beyond RAS, I continued to work on lightweight virtualization solutions based on containers with multiple teams for kernel and hardware, microservices and serverless teams at Facebook. Working with dozens of people on a variety of projects has been a very rewarding experience and a unique learning opportunity.
Q: What is it like to do research on Facebook?
DS: Good research is always motivated by asking the right questions. What problems should the next generation of software and hardware systems solve? How should we maximize our research impact? Facebook is an ideal place to answer these questions and gain valuable experience. The opportunity to do great work and write strong technical papers at high level conferences is possible. This is evident when you look at the strong research publications of several teams and projects on Facebook.
Of course, that doesn’t mean that there weren’t any challenges. One challenge I had to overcome was staying a little outside of my research comfort zone in order to grow. I believe the best approach to addressing these challenges is to network with people with expertise in multiple fields and to work together to solve problems at all levels.
Facebook encourages collaboration and direct communication between different teams and projects. During my visit, I had the opportunity to connect with people from many teams and organizations across distributed systems, operating systems, hardware, machine learning, security, and others. It’s very exciting to talk to people who work on very different projects and products. Such conversations could lead to long-term collaborations or further help to connect the dots and create an understanding of the bigger picture.
Q: What advice would you give university researchers who want to become visiting scholars on Facebook?
DS: I believe the best approach to becoming a visiting scholar is to network: attend conferences and contact people who have visited or worked on Facebook in an area that interests you. This is the best way to get a first impression of the experience and maybe get started. My experience began when I met Tianyin Xu, who also spent a year as a visiting scholar on Facebook.
As a visiting scholar, one of the most valuable pieces of advice from my Facebook colleagues Kaushik Veeraraghavan and CQ Tang was not to feel constrained by research areas. My main research interests are abstractions and interfaces that connect hardware with operating systems. At Facebook, I had the chance to connect with layers higher up in the stack and work on bridging hardware with distributed systems and creating abstractions for hardware and container management on a regional level.
Just go for it. Facebook provides an environment to pursue your research interests with leading experts and do impactful research. It’s a unique learning experience. After joining Facebook, it became clear that the possibilities are endless, so my main advice to visiting scholars is to get too involved – but be controllable. Solving real-world problems on the scale of Facebook takes a lot of time and effort, and taking on multiple projects has given me more opportunities to work on and explore different directions with multiple teams. That helped me to see the big picture and to connect all parts of the research.
Q: In your opinion, what are the next big challenges in systems research that industry and science could tackle together?
DS: I believe computer systems are undergoing radical change, driven by stringent security requirements and an unprecedented growth in data and users. This change has proven to be a breach of abstraction. Current hardware and operating system abstractions were developed at a time when we had minimal security threats, scarce computing and storage resources, and a limited number of users. These assumptions are not representative of today’s computing landscape. In this new era of computing, there is an urgent need to rethink the synergy between the operating system and hardware layers. The collaboration between industry and academia will be critical in building the systems of tomorrow that will power the world’s data centers.