Hi, Yes, you are right, but only ParquetMROutput miss the merging cuboid data step, because it will been done in SparkBatchMergeJobBuilder2 when using spark engine. Would you like to create a jira and fix it? thank you.
On Tue, Jun 11, 2019 at 6:22 PM pannenggang <[email protected]> wrote: > Hi > > I found a branch named “kylin-on-parquet” here: > https://github.com/apache/kylin/tree/kylin-on-parquet. I am working on a > little poc based on it. I have a question about the function > “getBatchMergeOutputSide” in class org.apache.kylin.storage.parquet.steps. > ParquetMROutput(similarly ParquetSparkOutput), basically, I think there is > no actually merge steps here, why is that? No need to do a merge step here? > > Best Regards >
