Google Cloud Datastore

• Datastore eliminates the need for a DBA. Google takes care of:
  • high availability
  • scalability
  • redundancy
  • replication
  • sharding
• schemaless database

Datastore

Cloud Datastore (opens in a new tab)

• NoSQL store for billions of rows
• schemaless access - no need to think about underlying data structure
• local development tools
• automatic scaling and fully managed
• built-in redundancy
• support for ACID transactions
• includes a free daily quota
• access from anywhere via REST API

Kinds, Entities, Properties, and Keys

Kind

• categorises entities for querying purposes
• corresponds to a table in a relational DB

Entity

• data object containing properties
• corresponds to a row in a relational DB (and document in MongoDB)

Property

• data values within an entity
• corresponds to a field in a relational DB

Key

• identifier for the entity
• corresponds to a primary key in a relational DB

Entity Groups

• entities can be organised into groups
• can include different types of entities
• an entity group is a set of entities connected through an ancestry, forming a heirarchical structure
• sequence of entities from parent to child to a specific entity is known as its ancestor path
• entity groups are required for achieving:
  • strong consistency, and
  • for participating in transactions

Datastore Development

• datastore client libraries:
  • on GAE: Java, Python, Go, Node
  • not on GAE: whatever you're willing to build / can find
• queries
  • supports filters, sorts, and special queries
  • familiar SQL-like language - GQL
  • most queries are eventually consistent
• indexes
  • all queries served by pre-built indexes
  • every property is automatically indexed
  • can define custom indexes
  • performance scales with size of result set, not the size of the dataset

Indexes are store using Google's BigTable.