How’s this for a challenge? Create a computer based on technology that currently does not exist; analyse and store more data than what is on the daily Internet; and work on a billion dollar project with the formable name of Square Kilometer Array (SKA) that has drawn worldwide attention. And your goal: answer the biggest scientific question that have puzzled humankind for hundreds of years — “Why are we here?”
|Yusik Kim (left) and Rik Jongerius
at the Westerbork telescope in Netherlands.
At the official launch of the ASTRON and IBM Center for Exascale Technology in Dwingeloo, Netherlands, where they work, we gave them a sanity check with a few questions.
Editor’s Note: Join Rik and Yusik in a live virtual recruiting event on 26 March. Click for details on how to participate and maybe even apply to join them.
What specific challenges are you each trying to solve with DOME?
Yusik Kim: The SKA will capture big data via the antennas — roughly twice the volume on the Internet today. This is one of seven challenges for the project, which is to figure how to store this data and then how can we make it accessible to thousands of astronomers around the world. On top of this, it needs to be energy efficient and affordable. So, we have to understand how we can leverage today’s storage technology, like tape and hard disks, yet make it general enough to include future technologies like phase change memory. The biggest mistake would be to constrain ourselves based on what is only available today.
Rik Jongerius: One of the other challenges or work packages, as we call them, is algorithms and machines and here we want to achieve two goals. First, once we have a system design we want to verify that is is feasible. Can it actually do the computations required? And can it do this with as little energy as possible?
To the second challenge: we know we can’t build the SKA based on current technology, therefore we need to narrow our focus areas to the right machines running the right algorithms. It may sound simple enough, but there are dozens of approaches and if you know any scientist, you know that we all have our own opinion.
|A closed dense aperture array that will be used in SKA|
What do you find the most fascinating about the DOME project (DOME is the project name for the development of the IT roadmap by ASTRON and IBM)?
YK: The most fascinating aspect is the sheer size of DOME. I have never been involved in a project that has such enormous scope and size, not to mention an operational lifetime of 50 years. I’ve been here two months, and I’m not sure if it will ever really hit me. Since we each have our own work packages it helps to break it down, but its not a trivial exercise to understand how all seven combine together.
RJ: From a technical perspective there are so many unknowns, which is unique. But what I find most interesting is that this project spans so many disciplines. As a computer architect, I enjoy dusting off my old physics books to understand how a telescope works because I haven’t thought about astronomy in years. I also appreciate the scale of the people involved from so many countries and organisations. The SKA is massive collaboration.
How did you find out about the job?
YK: I received my degree in 2009 from University of California-Berkeley and then studied for two years in France at the National Institute for Research in Computer Science and Control on my post-doc, focused on studying access patterns for particle physics data. It was interesting, and then I found this job opening on the IBM website which combined my access patterns research with astronomy (which was my childhood dream). So this job is perfect because it combines my passions. It really is a unique opportunity.
RJ: I just finished my masters thesis during an internship at IBM Research – Zurich. I guess they thought well of my work because they offered me to stay on as a Pre-Doc. It’s a great project because I feel apart of something big; something that can rewrite history.
How confident are you that DOME will succeed?
YK: Based on my two months of experience, I can only say that if we progress at our current rate we will have good results at the end of DOME. It’s a bit too early for predictions, but I am confident.
RJ: I am pretty confident as well, but I can’t forget that this is a very challenging project. To be successful we can’t be afraid to take on some risks.