Select three NoSQL databases that meet these requirements: ingest telemetry from millions of IoT devices, ~100 TB/year growth, ~100 attributes per record, no ACID needed, require high availability, low latency, and the ability to query individual fields. (Choose three.)

HBase, MongoDB, and Cassandra are all NoSQL databases well-suited for this scenario. HBase is a column-oriented database built on HDFS, offering high scalability and low-latency access for large datasets, ideal for time-series IoT data. MongoDB is a document-oriented database known for its flexibility, high availability, and ability to handle semi-structured data with many attributes, and it supports querying individual fields. Cassandra is a wide-column store designed for high availability and linear scalability across many nodes, making it excellent for high-ingest, low-latency applications like IoT telemetry. Redis is an in-memory data store, primarily for caching and real-time analytics, but its persistence model and storage capacity are not ideal for 100 TB/year growth. MySQL is a relational database, which typically struggles with the scale and schema flexibility required for millions of IoT devices and high ingest rates. HDFS with Hive is a batch processing solution, not designed for low-latency queries or individual record access.