Thats right, hard power switch.
They had soft power on some devices (DS2000, DG1022). But now they've changed everything to mains switches (MSO5000, DG800, DP832), which I like.
I haven't heard of any corruption issues with MSO5000, so hard switch is fine if done right.
However, I've seen it done wrong, causing many issues. But in that case the device was writing regularly to the NAND, I don't believe MSO would need to do that.
Normally, this is bad for filesystems, for electronics, it does not matter. In ye old days, harddisks used to 'park' their heads just before shutdown, but I think these days, they do this even without power (magnets and springs? parked by default?)
The problem exists really only for filesystems as you may have open files, be half-way through a write to storage and them BAM, power gone, corrupted file system.
The recent Rigol's MSO NAND architecture is very resilient. It's very hard to brick.
Can you explain what you mean here?
They use 'standard' MLC chips with standard drivers, nothing resilient there. They do use a 'read-only' filesystem with few writes, which is ok, but they still allow writes to the filesystem (good for us, but really bad for the device). They should have used squashfs with an overlay or something to be more robust