Large experiment rendering in Portal wizard/cluster picker
When swapping in large OpenStack experiments during the Capnet OSDI paper (and by large I mean 25-ish nodes), I noticed that the wizard process can get blocked for quite a long time (i.e., 1-2 minutes) -- it really ate up my old desktop CPUs . I went to Jon and he fired up some profiling stuff and it seemed to be largely in Jacks rendering code.
25 nodes doesn't really seem like a large experiment. So probably something needs to be done here, although Jon explained to me that it is quite complicated!