We managed to ask him few questions while he was on his way to the airport. You can also hear his talk during the session 6A, track 2 on Wednesday 2nd September (see the programme for details).
Q1. Who are you?
AU: I’m Alexandru Uta, born in Bucharest, Romania. I did my Bachelor in Computer Science in the same city, at the University of Bucharest, Faculty of Mathematics & Computer Science.
For my Master’s degree I moved to Amsterdam where I studied High Performance Distributed Computing at the Vrije Universiteit.
In the same place, I’m currently pursuing my PhD degree under the supervision of dr. ing. habil. Thilo Kielmann and prof. dr. Henri Bal. I’m currently working on an in-memory runtime distributed file system that accelerates I/O operations for scientific workflows.
Q2. How did you originally get involved in the eScience? What were your expectations then?
AU: I started to get involved with eScience applications when I started my PhD project. Initially, I actually did not know that, but when digging deeper into other applications that can benefit from my storage system I realized that it’s a promising data storage infrastructure for eScience applications (e.g. astronomy, bioinformatics etc.).
Q3. What is the most interesting issue you’re working right now?
AU: Initially, my work focussed on showing how optimizing for data-locality in runtime in-memory distributed file systems for scientific workflows can sometimes hurt performance of such applications. Now I’m turning the page a bit and I’m trying to see how locality-aware systems can be improved to take (more) advantage of the underlying hardware platforms.
Q4. You end up in an elevator with a representative of a funding agency – what will you pitch and how?
AU: Nowadays the volumes of data to be analyzed by applications have exploded and storage systems are lagging behind the computational power of current computing systems and don’t bring data fast enough to the application. The in-memory storage system I’m developing for scientific workflows bridges this gap and accelerates I/O for these applications. I believe that improving storage performance for such applications is extremely important for the scientists that develop these applications, as faster (and cheaper) execution of their experiments could drive their domain-specific research (astronomy, bioinformatics, earthquake modelling etc.) forward and enable further scientific breakthroughs.
Q5. What are your plans for the 2015 eScience conference? (e.g. talks or interest, meetings planned etc)
AU: I’m looking forward to attending Ewa Deelman’s keynote talk [9.00 – 10.20 AM CEST on Wednesday – see the programme for details] and various other conference talks that focus on data-intensive science and storage systems.
I’m also looking forward to finding more applications suitable to my storage system.
Q6. What would be your advise to someone starting at a university who wants to do eScience?
AU: I guess it probably depends on what area of eScience will be studied. Talking from my experience and for the field I’m currently working on, I’d advise a (future) student to dig deeper when studying (courses about) distributed systems and algorithms, parallel programming. The knowledge you gain from these topics can provide very important and powerful tools when tackling eScience problems.
Q7. What was the most important question we forgot to ask – and how would you answer it?
AU: Question: “What is the end goal of your current project?”
And the answer: I’d like to see domain-specific scientists adopting my storage system and using it successfully to make their experiments faster and cheaper to run.
Thank you very much for your time, bon voyage and welcome to Munich!