New professor wants to create space for the enormous data volumes of the future
The growth of data volumes cannot be sustained with the architecture of current computer systems. Philippe Bonnet, newly appointed Professor at the IT University of Copenhagen, is making data storage possible in a future with technologies that have not even been invented yet.
Written 8 November, 2016 08:26 by Jari Kickbusch
Every ten years, a new class of computer emerges: In the 70s it was the mainframe, in the 80s personal computers, the 90s the web, in the 00s smartphones and the cloud, and in the 2010s IoT (Internet of Things). What will a new class of computer integrating new forms of non-volatile memories look like? And how will this impact the design of data management systems? These are the questions newly appointed Professor, Philippe Bonnet, focuses on:
“When I was a kid, we were using audiotapes to store games. Then came floppy disks and hard drives. Today, we are using Solid State Drives (SSDs) based on NAND flash in our laptops. Tomorrow, other non-volatile memories will take over,” he says.
Sustainable data storage
As completely new forms of memories are emerging that challenge software designs that have been stable for decades, Professor Bonnet’s research focuses on rethinking the design of data management software on modern forms of storage hardware. A specific problem is that the exponential growth in data volume cannot be sustained with the architecture of current computer systems where data has to travel long distance.
In essence, the idea is to move processing into memory and storage rather than moving data from memory or storage to a central processor.
Philippe Bonnet, Professor, IT University of Copenhagen
“A solution is to make it possible to process data as close as possible to where it is stored. In essence, the idea is to move processing into memory and storage rather than moving data from memory or storage to a central processor. The concept of active disks and active memory, i.e., storage and memory devices that can be programmed, was discussed 20 years ago, but was never really developed because there was no need for it. The evolution of the hardware in recent years makes it reasonable, and possibly necessary, to envisage such solutions today,” he says.
Leading team in Europe
Philippe Bonnet’s ambition is that his group remains the leading European research team in the area. He wants to develop existing collaborations with the best universities in the USA and China so that the current and upcoming generations of MSc and PhD students have an optimal context for their studies. Furthermore, he stresses the importance of impact outside academia:
“In the context of the CLyDE project, funded by The Danish Council for Independent Research (DFF FTP), we prepared a couple of patent submissions, which are still pending. Our new research projects will build upon this Intellectual Property and develop it further. Our excellent contacts with leading companies such as DellEMC, NXP and Intel enable us to keep in touch with the latest developments in industry and possible paths towards the commercialisation of our Intellectual Property,” he ends.
Inaugural lecture: ’Will Near-Data Processing Change Everything?’ Friday, November 11 at 2-3 pm in Auditorium 4 at ITU.
Philippe Bonnet, Professor, email phbo@itu.dk
Vibeke Arildsen, Press Officer, phone 2555 0447, email viar@itu.dk