Will French steps down as ACCRE’s Director of Research Computing Operations
Will French, ACCRE’s director of research computing operations, has left Vanderbilt after nearly five years of service to ACCRE. Will came to Vanderbilt in 2008 as a graduate student in chemical and biomolecular engineering; he joined us shortly after completing his PhD and eventually became co-director of ACCRE in 2017.
Anyone who has used ACCRE in the past few years has received more than a few emails from Will about scheduled downtime on the cluster or other cluster updates. Will has overseen many of ACCRE’s changes, including its transitions to SLURM, Lmod and CentOS 7, and has guided us through many of its difficult challenges. He has started the Pizza and Programming lecture series, in which faculty and staff are invited to give lectures on high-performance computing topics, and has also taught the Scientific Computing course on parallel programming and high-performance computing. Thanks to his guidance, we were also able to launch this very website last year.
“I remember when I was teaching Will in one of my HPC courses,” says Alan Tackett, ACCRE’s technical director. “I was impressed by how much he has grown and I enjoyed working with him over the past five years. He did an amazing job.”
Will is about to join Atiba Software as a Senior Software Engineer this month. Before he left, I had a chance to ask him a few questions looking back at his career.
What accomplishments at ACCRE are you most proud of?
I’m most proud of the improvements we have made as a team – whether it be our parallel filesystem, network, job scheduler, software stack, or deployment of interactive environments. I’m also proud of the impact of the HPC I taught each spring over the last five years.
Which of ACCRE’s research groups has interested you the most?
It’s hard to pick just one, and my answer would depend on when you asked! Most recently, I have enjoyed working with a group from the Division of Epidemiology in VUMC that stores and analyzes whole genome sequencing data on ACCRE resources. This group works with massive amounts (PB scale) of data, and ACCRE staff have been working closely with them to design some data pipelining tools that make optimal use of GPFS, LStore, our new Black Pearl tape archive system, and local storage on compute nodes at various points of the analysis. This work has relieved some of the stress from GPFS while also hardening ACCRE’s in-house LStore system.
What has been your biggest challenge?
Maintaining a large shared environment that’s used by over 1k researchers who each have unique usage patterns and needs is quite the challenge. We’ve experienced a large amount of growth (mostly in data volume and access patterns) since I’ve been here, and some growing pains as a result, but I think we’ve made huge improvements in some key areas along the way.