Congrats! That is Redis-on-Xeon territory! https://redis.io/topics/benchmarks

Though I assume that, like Redis, a reboot would lose acked writes?

An SSD can't save a write in a microsecond, its latency is at least an order of magnitude higher, right?

Enterprise drives usually have ultra caps with enough energy to power the writeback of the writebuffer to stable storage in case of power loss, thus your write latency is just the time it takes to hit the memory buffer, typically dominated by the PCIe transfer, that is, a few µs (unless of course, the buffer has filled but that means hitting peak write bandwidth).

Consumer drives typically don't pay for this, but instead good drives buffer writes in SLC flash which has lower latency than MLC.

ADDENDUM: funny thing is that you have to actively manage the energy in the caps; the writebuffer must never have more data than you have energy to write back. This becomes especially important on power-up where the empty cap is charging.

Yes, unless you did a journal flush. Default journal flush interval is 1 second because of hard drives, but on a good SSD it can be turned up to around 10 ms and still maintain those numbers.

NAND program operations are in the order of O(100us) to O(1ms).

I see how you made that mistake now. SSDs typically have many (eg 10s-100s) program operations happening in parallel. They also pack each write into something around 16kb or larger units so it's possible multiple b+ op are in each program.