Lustre File System
In 1995 I started to work on distributed file systems, and - unexpectedly - in 1999 an opportunity emerged to design a new cluster file system which I named Lustre. Guided almost entirely by the requirements of the largest US government laboratories I led its architecture until 2008. Hundreds of individuals and hundreds of companies have made important contributions to it. It is now a cornerstone of scientific computing as one of the most used HPC software packages.
EC exa-scale strategy
From 2010-2012 I served on an expert panel to devise a strategy for the EC vis-a-vis exa-scale computing. The discussions were fascinating and resulted in the foundation of the European Technology Platform (ETP). However, I still don't know how one would decide what to apportion to infrastructure and services vs. research for fundamental progress.
The square kilometer array radio telescope needs seriously data intensive computing both to convert the input from the antennas into usable scientific data and to subsequently search that data. I work with the SDP project in Cambridge to develop a novel exa-scale compute approach to these problems, as well as advising the SDP project on other matters.
Few companies understand better than FABERNOVEL what makes technologies, teams and companies succeed (just have a look at their stunning presentations on Slideshare). Acting as an advisor it is a pleasure to formulate strategy and analyze industry trends with their leadership team. This eminent boutique firm is headquartered in Paris with offices in San Francisco.
At short notice, I visited Cape Town late 2013 and gave a keynote talk on HPC Data and Big Data. A relationship with CHPC and other organizations in the Department of Science and Technology emerged, and we are now defining big data analytics projects, creating opportunities for spin-out companies and formulating a differentiating role for South African super-computing.
Exa-scale I/O workgroup
In 2011 I started an open effort to analyze what an exa-scale I/O system should offer. Representatives from some 50 major labs, universities and companies joined the discussions. An architecture emerged, but in 2013 I withdrew from the group, because a nagging question remained: have the major players in the cloud already successfully addressed these problems with domain specific solutions?
About Peter Braam
Dr. Peter Braam is a multidisciplinary innovator covering computing, data and science. Research organizations, governments and enterprises engage with him to create concrete, focused visions and co-design solutions. As a recognized global leader in computing, without fear to be wrong or incomplete, he thinks systematically to advance technology frontiers.
Peter personally works on finding technical solutions for select projects. Seeing opportunities to apply his knowledge of computing to radio astronomy, he is currently working with Cambridge University on the SKA telescope Science Data Processor. He also consults with a few organizations and institutions on their strategy, for example with South Africa to further computing and big data efforts. Recently he architected a new exa-scale I/O system and brainstormed with the European Commission on the exa-scale HPC Strategy and the London Mathematical Society on the advancement of mathematics in the UK.
Often called “a big thinker attracted to complex problems,” one of Peter’s greatest strengths is to articulate unexplored avenues that provide the most opportunity for innovation and long-term impact. He passionately shares his and others’ insights in keynote talks at conferences and in white papers, and leaders of scientific and technological communities enjoy brainstorming with him in private discussions.
Having started his career as a scientist, Peter was a senior academic at Oxford University where he studied, researched and published with world leading mathematicians and physicists. In 1990, he revived his love for computing first in academia at Oxford and Carnegie Mellon, then as an entrepreneur. Peter focused on distributed storage systems and his noteworthy Lustre file system, which is among the most used operating system software for large-scale supercomputing. With its high-performance capabilities and open licensing, more than half of the Top500 supercomputers in the world use Lustre. Since 2008, he has mostly been working on parallel computing, programming languages and relationships with mathematics.
Peter founded and ran five startups — the first four of which were acquired — which taught him how to execute corporate strategy as a CEO. In larger companies, such as Red Hat, SUN and Xyratex, he participated in setting executive strategy and led R&D for products using state-of-the-art engineering processes.
When not working and studying, you might find Peter with his family, climbing and skiing in the mountains, enjoying classical music, or appreciating art and design.
I give talks about technology that do not disappoint, and universally the word "refreshing" has been used. I try to be objective, tune the content carefully to the audience and usually present both surprising content and statements that make people laugh. Please get in touch if you'd like me to speak. Speaker bio (PDF)