CockroachDB

View Current Viewing Revision #4 from 04/27/2018 10:52 p.m.

CockroachDB is a scalable, fault-tolerant, SQL database built on a transactional and strongly-consistent key-value store. It is backed by RocksDB and uses distributed consensus algorithm to ensure consistency, it is inspired by Spanner wait commit to implement serializable. It is currently in beta. (Because CockroachDB is rapidly changing, so many findings are based on design document, outdated documentation or available code.)

History

CockroachDB is inspired by Google’s Spanner and F1 to address fault tolerance and distributed consistency.

CockroachDB supports snapshot isolation (SI) and serializable snapshot isolation (SSI). It is implemented using RocksDB snapshot capability.

Views

Not Supported

Query Execution

Tuple-at-a-Time Model

The query execution is similar to stream processing. When executing a query, the execution plan is transformed into a DAG, a gateway node orchestrates participating nodes begin processing.

Concurrency Control

Multi-version Concurrency Control (MVCC)

Transactions in CockroachDB is implemented

Joins

Hash Join

Because CockroachDB is backed by key value store. There are two ways of join in CockroachDB, join-by-lookup and stream joins. Join-by-lookup iterate through one table and lookup for it in the other table. Stream joins is an execution plan based on hashing.

Storage Model

Custom

CockroachDB stores its data in the distributed key-value store.

System Architecture

Shared-Nothing

CockroachDB has two layers, the SQL layer and the storage layer. The SQL layer sits on top of the transactional and strongly-consistent distributed key-value store. In the key-value store, the key ranges are divided and stored in RocksDB and replicated across cluster. It exports structured data API of relational concepts. It exports standard SQL interface at the SQL layer. The SQL layer translates SQL statements into calls to structured data API.