A data scientist needs near-real-time SQL querying over a stream of GZIP-compressed files. Which option provides the lowest-latency way to query the stream using SQL?

Amazon Kinesis Data Analytics (KDA) is designed for near-real-time SQL querying over streaming data. It can directly ingest data from Kinesis Data Streams (which the GZIP files would likely be fed into) and apply SQL queries, providing the lowest latency for this use case. A Lambda transform can handle decompression if KDA's built-in capabilities don't suffice. AWS Glue is primarily an ETL service for batch processing and data cataloging, not optimized for near-real-time streaming SQL queries. Kinesis Client Library (KCL) applications require custom code for processing and then writing to another service like Elasticsearch, adding complexity and potential latency compared to KDA's direct SQL capabilities. Kinesis Data Firehose delivers data to destinations like S3, which is suitable for later querying but not for near-real-time SQL querying directly on the stream.