What is the best way to organize the process of working with photos?

Question

NAS

Good day everyone!
Please share your experience or suggest possible solutions.
At home, a fleet of several machines:
Monoblock on Core2Duo, 7GB RAM
System unit on i5, 16GB RAM
Self-assembled NAS on Atom
Notebooks on i5 and i7, 8GB Ram
Everything is connected via a gigabit switch. NAS is used to store photos (~50mb RAW), with remote access to it. In general, everything spins fast enough, but only if you work from one machine. One has only to connect from the second, or copy information at the same time, everything starts to slow down a lot.
Task:
After running the image processing process on the monoblock, connect from it via RDP to any free machine and continue working with the next material.
What will be the "bottlenecks" with this method?
Whether there is a sense in productive hdd on the server (for example WD RE)?
Are there any thoughts on the topic?
The most important thing is to work on one machine using the resources of others.

Answer 1

The most stupid and obvious thing to do is to understand why it slows down, there is not enough performance of the screw on us, or the processor or memory, or everything rests on the network and nothing but expanding it will help.
All advice without elementary diagnostics that you need to spend five minutes on is groundless and not worthy of attention.

Answer 2

Is your self-built NAS capable of iSCSI? Put FreeNAS there, it definitely knows how to be a target, and Windows has an initiator. Then it would be nice to see what kind of disks you have there. WD RE makes sense, it also makes sense to add a 128/256 gigabyte SSD drive for caching as part of Intel Rapid Storage technology, for example.
You can transfer the network to 10G, respectively change the network cards and the switch. By the way, about the latter, is it a switch or is it a hub?
Well, about the processing of photos. It all depends on what tasks you want to parallelize. Let's say batch processing like Auto Adjustment -> Resize -> Sharp -> Save or something more complicated?

Answer 3

In fact, the bottleneck here is always the disk. Processing programs will capture RAW files.
I'm afraid that the only reasonable option is to make a constantly updated mirror (more precisely, a clone) of the original files, either through RAID 0 or using some kind of backup utility, and work with the mirror (a similar technology is used, AFAIK, in a database like MySQL when to split queries).
Those. I think that without some form of physical clustering of information, all processes will still go very slowly, and the network will not help here.