Amazon Keyspaces now supports logged batches for atomic, multi-statement operations

Amazon Keyspaces now supports logged batches, enabling atomic, all-or-nothing multi-statement write operations for Apache Cassandra-compatible workloads.

• Group multiple INSERT, UPDATE, DELETE statements into single atomic units that succeed completely or fail entirely
• Maintains data consistency across multiple partitions and tables
• Simplifies migration from self-managed Cassandra to Amazon Keyspaces
• Supports conditional logic (IF EXISTS, IF NOT EXISTS) and advanced data types (UDTs, static columns)
• Up to 100 operations per batch (50 with static columns); 4 MB max payload
• Uses lightweight locking mechanism for isolation; differs from Cassandra's last-write-wins approach
• No additional cost; pay only for consumed capacity units
• Limitations: no IN operator, range operators, or counter updates; LOCAL_QUORUM consistency only
• Multi-Region: atomicity in source region only; replicated writes are asynchronous

Logged batches provide consistency guarantees for denormalized data across multiple tables, though with slight latency trade-offs compared to individual writes.