[
https://issues.apache.org/jira/browse/KYLIN-3552?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16611181#comment-16611181
]
Julian Hyde commented on KYLIN-3552:
------------------------------------
I suggest that your consider using Calcite. Calcite makes everything look like
JDBC. Therefore Kylin can continue to use JDBC as your API. Calcite already has
adapters for things like Mongo, Elasticsearch and files.
The difference between a simple adapter and a complex adapter is whether you
are able to push down operations. A simple adapter has an API for data, whereas
a complex adapter speaks algebra. Before long you will need a complex adapter
and you will end up building something as complex as Calcite.
> Data Source SDK to ingest data from different JDBC sources
> ----------------------------------------------------------
>
> Key: KYLIN-3552
> URL: https://issues.apache.org/jira/browse/KYLIN-3552
> Project: Kylin
> Issue Type: New Feature
> Components: Query Engine, RDBMS Source
> Reporter: youcheng.zhang
> Assignee: youcheng.zhang
> Priority: Major
> Fix For: v2.6.0
>
>
> Kylin already supports JDBC source, but it takes much efforts to develop an
> implementation to a new source engine, like supporting metadata sync, cube
> build and query pushdown. It’s mainly because the SQL dialects and jdbc
> implementations between source engines are quite different.
> We propose to develop a data source sdk, which provides APIs to help
> developers handle these dialect differences and easily implement an adaptor
> for one source engine. With this SDK, users can achieve followings from a
> JDBC source:
> # synchronize metadata and data from jdbc source.
> # Build cube from jdbc source.
> # query pushdown to jdbc source engine when cube is unmatched.
> The SDK will define some interfaces of adaptors. To implement a new source,
> developers only need to implement an adaptor for the source.
> To keep compatibly, we will keep the existing JDBC source module, and
> implement a new JDBC source module as the caller of the SDK. Besides, we will
> also provide a pushdown implementation class as the caller of SDK for
> pushdown scenarios.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
