Inline large value at the highest level?

[DorianZheng]

Can we inline large value at the highest level for improving range scan performance at the cost of some potential write performance regression.

[yiwu-arbug]

Interesting idea! You can also take a look at the level_merge option that we are adding. The two idea have some similarities, but the level_merge one is more complicated, have lower expected write-amp (inline-bottom-level write-amp: 1(WAL)+1(flush)+10 (bottom level), level merge write-amp: 1(WAL)+1(flush) + ~3 (1 for each of bottom two levels and addition ~1 for the last level of so call "range merge")), but probably with not so perfect scan performance.