Some drawings and pictures and such need to go up here.
When I worked at Argonne National Laboratory, we used GPFS rather extensively. My relationship with the filesystem was pretty bipolar: when it worked, I really
liked it. And when it had problems, I really
hated it. Eventually we concluded that sometimes it just needed a plunger to clean out the tubes, so I drew one up
to help the cause.
When I was a grad student, I wrote a cluster health visualization tool modeled after a fishtank
. The project has since been passed from hand to hand among other grad students at UNL, but I've always been fond of the little fish that I created. In early 2011, I finally decided to recreate him with Blender
Science per hour
For whatever reason, high performance computing managers like to refer to "science" as a concrete thing: "We're getting a lot of science done on that cluster." At some point I decided we needed a little meter to show just how much science we were getting done at a given moment. Several kilosciences per hour
, it turns out.