In the world of PostgreSQL-compatible databases, a newcomer is YugabyteDB. Same protocol, similar open-source license, re-using Postgres query layer but on a distributed storage plugged in place of heap tables and btree indexes. We will explain the reason for it, why it cannot be just an extension, and why it re-uses PostgreSQL. Similarities and differences, with an overview of the underlying technology: LSM Trees, distributed storage, logical clock, consensus protocols.

The goal is to answer the questions like:
- why another database instead of an extension?
- when can it can substitute to standard PostgreSQL, and why?
- when it is not a good choice and PostgreSQL is the best fit?
- what is distributed vs. sharding when it comes to scale-out?