Batch API¶

The Batch class allows you to execute multiple statements atomically or as a group for better performance.

Batch¶

Constructor¶

from rsylla import Batch

batch = Batch("logged")  # or "unlogged" or "counter"

Parameters:

• batch_type - Type of batch: "logged", "unlogged", or "counter"

Raises: ValueError for invalid batch type

Batch Types¶

Methods¶

append_statement(query: str) -> None¶

Add a CQL statement to the batch.

batch = Batch("logged")
batch.append_statement("INSERT INTO users (id, name) VALUES (?, ?)")
batch.append_statement("INSERT INTO users (id, name) VALUES (?, ?)")

append_query(query: Query) -> None¶

Add a Query object to the batch.

from rsylla import Query

query = Query("INSERT INTO users (id, name) VALUES (?, ?)")
batch.append_query(query)

append_prepared(prepared: PreparedStatement) -> None¶

Add a prepared statement to the batch.

prepared = await session.prepare("INSERT INTO users (id, name) VALUES (?, ?)")
batch.append_prepared(prepared)

with_consistency(consistency: str) -> Batch¶

Set the consistency level for the batch.

batch = Batch("logged").with_consistency("QUORUM")

with_serial_consistency(serial_consistency: str) -> Batch¶

Set the serial consistency level (for LWT in batch).

batch = batch.with_serial_consistency("SERIAL")

with_timestamp(timestamp: int) -> Batch¶

Set a timestamp for all statements in the batch.

batch = batch.with_timestamp(1234567890000)

with_tracing(tracing: bool) -> Batch¶

Enable or disable tracing.

batch = batch.with_tracing(True)

is_idempotent() -> bool¶

Check if the batch is idempotent.

set_idempotent(idempotent: bool) -> None¶

Mark the batch as idempotent.

statements_count() -> int¶

Get the number of statements in the batch.

print(f"Batch has {batch.statements_count()} statements")

Executing Batches¶

Use session.batch() to execute:

result = await session.batch(batch, values)

Example¶

from rsylla import Batch

batch = Batch("logged")
batch.append_statement("INSERT INTO users (id, name) VALUES (?, ?)")
batch.append_statement("INSERT INTO users (id, name) VALUES (?, ?)")

result = await session.batch(batch, [
    {"id": 1, "name": "Alice"},
    {"id": 2, "name": "Bob"}
])

Best Practices¶

• Batch same partition - Group statements that affect the same partition key
• Keep batches small - Large batches can timeout or overwhelm coordinators
• Use unlogged when possible - If atomicity isn't required, unlogged is faster
• Don't batch across many partitions - This defeats the purpose and is slower