Compression

Dictionary Encoding Delta Encoding Run-Length Encoding Prefix Compression

OceanBase uses column compression for SSTable in disk. It implements diverse encoding algorithm and it will automatically choose the most suitable one for every column. It costs only half as much space as MySQL does.

Data Model

Relational

Storage Model

N-ary Storage Model (Row/Record)

Storage Architecture

Disk-oriented

OceanBase divides its systems into regions, every region has multiple zones, every zone has multiple ObServers, every ObServer has multiple partitions. Several Zones would store the same data to provide backup. Typically, there are four merge algorithms to choose from:

Storage Organization

Heaps

System Architecture

Shared-Nothing

Concurrency Control

Multi-version Concurrency Control (MVCC)

OceanBase adopts MVCC to do concurrency control. If the operation involves single partition or multiple partition on one ObServer, it will read the snapshot of that ObServer. If the operation involves partitions on multiple ObServer, it executes distributed snapshot read.

Stored Procedures

Not Supported

Query Interface

SQL

Checkpoints

Not Supported

Unlike tradition relational DBMS, OceanBase doesn't require and doesn't rely on checkpoint to ensure data consistency. There are two reasons for this: (1) OceanBase is deployed across zones with multi-copy data, so it can suffer from ObSingle server shut down without checkpoint. (2) In actual production, OceanBase stores the modification of data in MemTable in memory and merges that to SSTable in disk once everyday. This read-write isolation mechanism doesn't require frequent checkpoint. There are multiple merge strategy choice, this will be covered in Storage Architecture part.

Views

Materialized Views

OceanBase supports materialized view well.
Their first business, Taobao Favorites, is done perfectly by leveraging materialized views.