I haven’t tested it recently, but previously Nuke wasn’t most optimal about multiple read nodes sourcing exactly same file sequence. Each read was producing read operation which isn’t fastest or optimal. Meanwhile a lot of people really like copy/pasting read nodes instead of managing node graph properly. If you pickup their scripts it’s usually pretty slow.
To resolve this problem I created simple python script that scans node graph and whenever it finds duplicate read node, it will replace duplicate with a postage stamp node connected to a main read.
This will maintain all the connections, but will make sure read node only called once when needed.