Besides the use of LSM trees in RocksDB and leveldb-like databases, there is also the WiscKey approach (https://www.usenix.org/node/194425) that helps read/write amplification by keeping the LSM tree small and mostly removing the values to a separate value log. There's a pure Go implementation of the WiscKey approach used by dgraph: https://github.com/dgraph-io/badger
Seems like OReilly really goes out of their way to hide ALL books these days. When you go to the main site (https://www.oreilly.com/), where do you see anything related to books? All I see is "online learning", "blended courses", "conferences" and "ideas".
I'm a bit upset by this, because I've found the Safari experience terrible.
I found the build up to this through flavors of disk IO in parts 1 and 2 to be very useful to me. Basic stuff but because I had not done system programming since my college days it was a good refresher (and eye-opener).