On Fri, 2022-08-12 at 14:48 -0300, George N. White III wrote:
> Manually partitioning disks was an onerous task for sys admins in the bad old days. We didn't
> have UUID's, so had to be careful to keep track of which disk needed to be partitioned, write down
> the planned scheme, and carefully enter the parameters as text. Today's linux installers
> do a good job without the need for manual intervention unless your use case is far outside the
> norm.
Many the time I'd opened the box, written down the brand names and
sizes of the installed harddrives. It was easier when they were
different, and a right pain when a box had two or more identical
models.
I usually bought disks in bulk -- cheaper that way and I only had to
have a dos system with SCSI that could boot one vendor's diagnostics.
I retrospect I should have used the brand and serial no. to make my
own almost UUID. I did put a text file in the root of each drive with
notes on the partition structure, etc.
I haven't multibooted for eons (it has its own worlds of pain), but I
still like the idea of one drive for data, another for the system. It
makes OS updates so much easier to manage.
A separate data disk is a good strategy for a single OS. For multiboot
I need data to be available to all the OS's, so use a network drive
shared via samba.
The majority of my colleagues work for governments or universities
and are required to run windows, but use linux for large scale
processing. People who have the option mostly use macOS. ESA
and NASA have "mission critical" applications written in Java that
run on Windows, linux, and macOS. I need all 3 OS's in order to
investigate issues, and now need to add an Apple silicon version
of macOS.