Expand description
Account-partitioned, serializable state for Amazon Timestream.
Everything a Timestream caller can create lives here: databases, tables,
the ingested data points (a bounded in-memory buffer per table so Query
can read them back), scheduled queries, batch-load tasks, the per-account
query settings, and in-flight query ids (so CancelQuery resolves). Config
blobs (RetentionProperties, MagneticStoreWriteProperties, Schema, a
scheduled query’s schedule/notification/target config) are stored as their
already-output-valid wire JSON so a Describe* echoes exactly what its
Create* persisted.
Databases are keyed by name; tables and their record buffers by
"{database}\u{1}{table}" (see crate::shared::table_key); scheduled
queries and batch-load tasks by their minted id/ARN. Tags live in an
ARN-keyed side map so ListTagsForResource has a single source of truth.
Structs§
- Account
Settings - Per-account query settings.
- Batch
Load Task - A batch-load task.
- Database
- A Timestream database.
- Scheduled
Query - A scheduled query.
- Stored
Record - One ingested Timestream data point, normalized to nanoseconds so the query
engine can time-filter and format it uniformly regardless of the write-time
TimeUnit. - Table
- A Timestream table.
- Timestream
Data - Per-account Amazon Timestream state.
- Timestream
Snapshot
Constants§
- MAX_
RECORDS_ PER_ TABLE - The maximum number of ingested points retained per table. Timestream itself
keeps data by retention window; fakecloud keeps a bounded ring so the
process footprint stays flat under sustained
WriteRecords. - TIMESTREAM_
SNAPSHOT_ SCHEMA_ VERSION