“We can provide the scientific community and industry with the infrastructure they need to calculate large amounts of data quickly”, Dragi Kimovski, Assistant Professor at the Department of Information Technology, explains. In a recent conversation he told us what scientists and experts like him can offer the fields of medicine and physics as well as other areas.
Millions upon millions of health records are stored, as well as countless measurement results from the natural sciences and industry. For disciplines such as these, the challenge is to filter out information containing “meaning” from the huge amounts of data. At the same time, they do not want to wait forever, but rather they need rapid calculations. Here, the work of Dragi Kimovski and his colleagues is very helpful: They build parallel and distributed systems, including super-computers and cloud/edge systems, which collectively form the computing continuum required to accomplish such colossal tasks.
Kimovski points out two relevant aspects: On the one hand, he aims to facilitate high performance computing, with high efficiency and computing power for algorithms such as those used in meteorology or in simulations of the planetary system. On the other hand, cloud and edge computing is being refined with the primary aim of reducing the costs of cloud providers and increasing the speed of their services. Dragi Kimovski uses Netflix as an example: “We want to reduce the maintenance and structural costs of streaming providers, while at the same time maintaining a high-quality user experience.”
But what is really concealed behind the term “parallel and distributed systems”? Kimovski explains: “These are many interconnected servers. The super-computers and cloud systems we work with are multiple servers that only appear to be a single system.” Essentially, the common individual PCs in households could be interconnected and then, jointly, provide a large amount of computing power.
What is decisive, however, is who will calculate what in the future. Let’s take a self-propelled car driving at 100 km/h as an example. If an obstacle should get in the way of the vehicle, it would take far too long to send the data to a cloud or a supercomputer. Assuming a delay of 100 to 500 milliseconds – depending on the constellation – the accident would already have happened by now. “Time-critical services must therefore be brought closer to the applications. So, we will need to move computing power from the cloud to the edge of the network to support the Internet of Things application. So-called edge and fog computing forms the root layer of the computing continuum, which is handled by mini data servers located closer to the data sources”, Dragi Kimovski tells us in detail. Together with his colleagues, he wants to work on artificially intelligent algorithms that will simplify the decision of where and what is being computed – whether in the cloud or in the edge. Dragi Kimovski goes on to say: “If we want to “calculate” an urgent road situation, we will do this via edge computing. But if we want to learn something fundamental about the behaviour of vehicles, we have more time, but also need more volume. The cloud or a supercomputer is better suited for this purpose.”
Dragi Kimovski’s career path that led him to Klagenfurt was pretty steady. Originally from North Macedonia, he completed his PhD studies in Sofia (Bulgaria). Through networks related to parallel systems he met Radu Prodan, who heads the research group at the Department of Information Technology. He followed him to the University of Innsbruck and later to Klagenfurt. All told, Dragi Kimovski has been working in the academic world for 13 years now, and would like to stay there. Computer science offers many points of contact to industry and companies; ultimately, however, the university framework offers more freedom to work on specific issues. Dragi Kimovski has been interested in computer science since his school days, but he also understands that many first-year students are “a bit anxious about the subject, especially about mathematics”. But he is quick to offer reassurance: “Those who study with commitment and motivation will overcome hurdles.” He has noticed that the student commitment is more pronounced in Klagenfurt than elsewhere. He is particularly motivated by the fact that “computer science has a great influence on all other sciences. Without it, many insights could not be generated today.”