COMPARING APACHE HIVE TO APACHE IMPALA
In one of its blogs, HortonWorks shares interesting insight into Apache Hive with LLAP (Low Latency Analytical Processing). Apache Hive is easily the best SQL engine in the Hadoop ecosystem, with ACID, security, Spark access etc.
“Hive’s core execution engine and operators are highly optimized and designed to work end-to-end with the ORCFile columnar format and LLAP’s in-memory cache. Given this level of optimization, the majority of Hive’s performance improvements in HDP 2.6 come from more sophisticated cost-based optimization along with dynamic runtime features that minimize the overall work Hive must do to satisfy a query. You can say that with HDP 2.6, Hive is working smarter and not harder, re-using its own work or avoiding it altogether when possible.”