Skip to main content

Batch read settings

On a block-by-block basis, customize how it reads data from its upstream blocks.

Read specific chunks

Downstream blocks can control the volume of data it loads into memory from its upstream block’s output. The downstream block will only load the chunk from indigo_mountain where power equals 5.
1

Load data

2

Transform data

Input data types

Downstream blocks can control the strategy it implements when loading an upstream block’s output data into memory.

Batch

Read data in batches.

Generator

Batch generator framework to operate and process 1,000+ gigabytes (GB) of data without running out of memory.
1

Load data

2

Transform data

Reader

Invoke methods on the reader object directly.

Batch write settings

Customize how data is written when outputting data from a block.

Strategies

Explicitly set the chunking strategy when creating data partitions or batches of data. Example: create chunks by the value in column power

Items per chunk

Example: each chunk can only contain a maximum number of 7,777 items.

Number of chunks

Example: each chunk must have at least 1,337 items and the total number of chunks cannot exceed 143.

Byte size per check

Example: each chunk cannot exceed 100MB in size on disk.

Modes

Control how data is written.
  1. Append
  2. Replace