A fleet of Amazon EC2 instances ingests JSON data from on‑premises sources at up to 1 MB/s. When an EC2 instance reboots, in‑flight data is lost. The data science team needs near‑real‑time querying of ingested data. Which solution provides near‑real‑time querying, is scalable, and minimizes data loss?

Publishing data to Amazon Kinesis Data Streams ensures data durability and availability, preventing loss during EC2 reboots. Kinesis Data Streams supports near-real-time processing and is highly scalable. Kinesis Data Analytics can then directly query the streaming data in near real-time using SQL, fulfilling the data science team's requirement. Kinesis Data Firehose with Redshift is good for analytics but introduces latency for near-real-time querying of streaming data directly. EC2 instance store is ephemeral and doesn't prevent data loss. Kinesis Data Firehose to S3 with Athena is suitable for batch querying of historical data, not near-real-time streaming analytics. EBS and ElastiCache for Redis are not designed for persistent, scalable streaming data ingestion and near-real-time SQL querying of streams.