Skip to main content
Version: Next

Localpod Data Connector

The Localpod Data Connector enables setting up a parent/child relationship between datasets in the current Spicepod. This can be used for configuring multiple/tiered accelerations for a single dataset, and ensuring that the data is only downloaded once from the remote source. For example, you can use the localpod connector to create a child dataset that is accelerated in-memory, while the parent dataset is accelerated to a file.

The dataset created by the localpod connector will logically have the same data as the parent dataset.

Synchronized Refreshes​

The localpod connector supports synchronized refreshes, which ensures that the child dataset is refreshed from the same data as the parent dataset. Synchronized refreshes require the parent and child accelerations to use the same refresh mode, and only two pairings qualify: both refresh_mode: full (which is the default), or both refresh_mode: caching. A mismatched pair — a caching child under a full parent, for example — is not synchronized.

Synchronization is best-effort. When it cannot be established — mismatched refresh modes, a parent that is not accelerated, or a parent dataset that is not registered — the child silently falls back to refreshing on its own schedule, and the reason is logged only at DEBUG. At default log levels the INFO line below is the sole confirmation that synchronization took effect; its absence means the two datasets refresh independently.

When synchronization is enabled, the following logs will be emitted:

2024-10-28T15:45:24.220665Z  INFO runtime::datafusion: Localpod dataset test_local synchronizing refreshes with parent table test

Examples​

datasets:
  - from: postgres:cleaned_sales_data
    name: test
    params: ...
    acceleration:
      enabled: true # This dataset will be accelerated into a DuckDB file
      engine: duckdb
      mode: file
      refresh_check_interval: 10s
  - from: localpod:test
    name: test_local
    acceleration:
      enabled: true # This dataset accelerates the parent `test` dataset in memory and is synchronized with the parent

Multiple Children and Load Order​

Any number of localpod datasets can read from the same parent, and a localpod dataset can appear before its parent in the Spicepod. At startup, each child loads after the dataset it reads from. from accepts the parent's dataset name or its full table name, so localpod:taxi_trips and localpod:spice.public.taxi_trips read the same parent.

datasets:
  - from: localpod:taxi_trips
    name: taxi_trips_local
    acceleration:
      enabled: true
  - from: https://public-data.spiceai.org/taxi_sample.parquet
    name: taxi_trips
    acceleration:
      enabled: true
  - from: localpod:spice.public.taxi_trips
    name: taxi_trips_local_2
    acceleration:
      enabled: true

Hot Reload​

A localpod child binds to the table its parent has registered at the moment the child loads. When a hot reload changes the parent — an edited acceleration, a new refresh_sql, or the parent being removed and added back — the runtime reloads every localpod dataset that reads through it as well, parents first, so each child rebinds to the parent's new table. Chains follow the same rule transitively: a child whose parent is queued behind a load of its own waits for it, and the chain is loaded from its root down.

Cookbook​