This document discusses different approaches to running SQL queries over Hadoop data. It identifies 6 types of SQL on Hadoop technologies: 1) MapReduce-based batch queries, 2) connecting external databases to HDFS, 3) parallel query engines that pull data from HDFS, 4) using HDFS as storage for MPP databases, 5) running local databases on HDFS nodes, and 6) distributed native SQL engines on HDFS. The document advises that when choosing a SQL on Hadoop technology, considerations should include ANSI SQL compliance, distributed data-local processing, support for file formats and compression, optimized querying, scalability, and low latency/cost.