Core concepts
Records and schemas
Section titled “Records and schemas”Tables store typed records. Schemas group tables within a database. Choose stable primary keys and types that match how the application uses each value. Tables and schemas.
Transactions
Section titled “Transactions”A transaction groups database changes that must succeed or fail together. Keep transactions short. Retry the whole transaction after a retryable conflict; do not retry an external action blindly. Transactions.
Search and memory
Section titled “Search and memory”An embedding describes a record in vector form. A vector query finds nearby embeddings; relational filters narrow the candidate records. Generate embeddings consistently and use the same dimensions for stored vectors and queries. Search and memory.
Key-value
Section titled “Key-value”Keys are useful for short identifiers, counters and expiring data. Key-value access is a separate interface from ordinary table queries; check supported commands and atomicity boundaries before building coordination on it. Key-value.
Particles
Section titled “Particles”A Particle is custom code callable from SQL. Its inputs, outputs, privileges and execution limits form part of your application’s database contract. Particles.
Analytics
Section titled “Analytics”Aggregations answer questions about operational records without requiring an export to a separate system. Indexes, data size and query shape still affect response time. Analytics.
Interfaces and guarantees
Section titled “Interfaces and guarantees”One server offers several interfaces: SQL clients, the HTTP API, Redis-compatible clients and the native SDKs. Do not assume that transactions, authentication or values are interchangeable across separate connections; consult connection addresses and the reference for your client. Transactions, durability and limits describe the observable application contract rather than storage internals.
Related
Section titled “Related”This site uses Google Analytics cookies to count visits. Privacy · License