In addition to what Julian mentions, you might find Apache Drill interesting.
Much of what you want is already in Drill, plus you get a parallel execution engine. On Thu, Nov 19, 2015 at 4:08 AM, Julian Hyde <[email protected]> wrote: > Yes, this is exactly what Calcite is intended for. > > Those limitations are out of date. The JDBC adapter can push down all of > the standard relational operators (exceptions being NULLS FIRST/LAST, > GROUPING SETS and maybe window functions). > > You will need a schema adapter (i.e. a class that implements SchemaFactory) > for each of those kinds of data. > > > - Databases by JDBC > > Use JdbcSchema.Factory. This is so common that in a model.json you can just > say > > "type": "JDBC" > > > - REST Services > > We have a prototype web adapter; see > https://issues.apache.org/jira/browse/CALCITE-884. > > > - SOLR Search by SolrJ > > There is no SOLR (or Lucene) adapter currently. > > > - POJO > > You can build schemas on POJOs - see ReflectiveSchema - but there is no > SchemaFactory, so you need to instantiate the schema manually. > > Julian >
