# Separate Values from LSM-Tree
+ Separate large values to `BLOB` files to reduce the write amplification

+ An additional `I/O` may be required for each query
  + `LSM-Tree` is smaller so that it can be cached in memory easier

+ Range queries require random `I/O`
  + Leverage the parallel `I/O` capability of `SSD`
  + The difference between random `I/O` and sequential `I/O` is narrowed

```
~!@// Example@!~
key: 16B, value: 1KB
(40 x 16 + 1024) / (16 + 1024) = 1.6 
```
------------------------------------------------------------