Re:Apache Drill 1.0 – Schema-Free SQL Query Engine for Hadoo...
by eranation on Tuesday May 19, 2015 @10:03PM
(#9574286)
How does it compare with spark SQL? Does the difference lie in where the query runs? In spark it loads all data into memory first and then uses SQL to query it (it parses the query and converts it to a set of spark transformations, kind of like hive). Does Drill actually translate SQL into, for example, mongo db native queries and run it on the source data store? If so this is a whooping amazing feat...
[ Parent
| on HN ]
Re:Apache Drill 1.0 – Schema-Free SQL Query Engine for Hadoo...
by rxin on Tuesday May 19, 2015 @10:19PM
(#9574329)
Actually Spark SQL doesn't load everything into memory. Its data source API supports pushing predicates down, and if the data sources implement it, it even supports running aggregations and joins in the data sources!
Re:Apache Drill 1.0 – Schema-Free SQL Query Engine for Hadoo...
by MasterScrat on Wednesday May 20, 2015 @09:15AM
(#9576065)
I would really be interested in a comparison of BI tools performances using the JDBC/ODBC drivers from Spark vs Drill.