Interview with Daniel HAGIMONT, optimized systems to manage computing resources

We publish the second video of the series of interviews presenting research works of our different departments. Daniel HAGIMONT, teacher-researcher at IRIT in the Architecture, Systems, Networks (ASR) team SEPIA, talks about his research in the field of operating systems. What is at stake? Building systems capable of managing the complexity of current infrastructures and limiting energy consumption.

What is an operating system ?

It is a software layer that facilitates the use of hardware. The mission of an operating system (Windows, Linux) is to efficiently manage the resources and applications of a computer. Today, there is a strong demand focused on the power of the machines, linked to the development of new resource-intensive applications, especially in the fields of artificial intelligence and big data. Operating systems must therefore adapt to this evolution. The great revolution of the last few years has been to no longer build large computers but to gather a large amount of computing power by agglomerating a set of computers interconnected by a network. This corresponds to the birth of data centers, computer parks on which current research is focused.

What are the current challenges in this field of research ?

The birth of data centers poses particular problems. The first problem is to build systems powerful enough to efficiently manage the colossal computing resources of data centers. Some digital giants and large companies have their own data centers, but other players have opted for cloud computing, i.e. making servers available to clients (OVH for example). The second problem concerns the waste of energy resources and therefore the search for solutions to optimize the management of these resources. The objective of Daniel HAGIMONT’s research is, among other things, to find solutions for moving applications to a limited number of computers to limit energy consumption and optimize the overall performance of the infrastructure (load balancing). This is a kind of application compaction. Today, the teacher-researcher explains that it is necessary to make these computing resources available to users (edge computing). The challenge of research in this field is therefore to build systems capable of managing these complex infrastructures.