How do stream processing frameworks such as Kafka and Spark handle range queries while using distributed data storage?

Question

If the data is distributed across multiple nodes doesn\'t that mean a full scan is required for range queries ?