Hi @jdx-john -
Yes, with the .mrb option the file is saved in a temp directory and then zipped (if you change a .mrb extension to .zip you can look inside). The opposite happens on load of an mrb. If you make incremental changes and then re-save you end up with the potential of a lot of duplicated source data compared to the .mrml approach.
Also note that volumes get compressed on save by default and for large volumes that can be big. Then the zip (mrb) process re-runs compression of those which is also wasteful. If performance is a big issue for you it’s possible to bypass that.
You could also look at FUSE with either a tarfs backend to keep things organized an hopefully efficient without needing to change Slicer.