All drives are HDDs. I am backing up drive A on host A to drive B on host B and drive C on host C. The hosts are connected via a LAN with a star topology. I started two consecutive rsync processes without considering whether that would be too taxing for drive A. Is that even a thing? Like, mechanically. Is the needle needlessly (sorry) jumping back and forth more than what's good for it? Maybe I'm being too empathetic towards the machine. Please enlighten me. :3

There will be limitations throughout your network; LAN speed, write speed on B, write speed on C, etc. these factors may slow down the reads on A such that it is never reading at full speed no matter what.

If you took a simpler example, say old SATA disk A being read the. Getting written to NVME disk B and C on the same machine, you are much more likely to start hitting HDD caps on the reading device, where delays from seeking may occur. Even then, your OS should be providing some level of "smoothing". It could be either reading whole files or large blocks of data from the HDD at any one time, such that it will try to make the most of a read at a location while not bottlenecking the sending of data to both write devices. Not to mention that rsync may also be smart enough to cache data between the two outputs such that two readys don't need to happen.

With all that said.... Does it matter? Not really. The OS will try it's best to get data out of A as fast as possible. The network will try to get data distributed in the same way. Disk A may be working hard and may do some jumps and read/writes, but it won't intentionally break itself.

Modern storage systems and protocols have device performance and longevity in mind, too.

A lot of really smart people have thought long and hard about the problems and limitations of the components that computers run on and how they communicate, and the longevity of the components is a testament to that collective effort.

The OS will normally buffer writes in RAM when an application does a write, then when it has a bunch, do them all at once, so it'll do a big chunk of one, then the other.

Also for some unusual filesystems, like, say it's a log-structured filesystem, the actual data may just be more-or-less written forward for both of them.