Hi. I think you can write a udf[1] to process some fields and then insert into the sink.
Best. Shengkai [1] https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/functions/udfs/ <pod...@gmx.com> 于2022年9月15日周四 22:10写道： > What's the most effective way (performance) to update big no of rows? > Sure this will be probably "INSERT INTO table (column1) SELECT column1 > FROM ...". Anyway, I do not see any "UPDATE" in Flink? > But sometimes SQL is not enough. > Suppose I have code: > > TableResult tableResult1 = tEnv.executeSql("SELECT * FROM SomeTable"); > try (org.apache.flink.util.CloseableIterator<Row> it = > tableResult1.collect()) { > while(it.hasNext()) { > Row row = it.next(); > //Treat row: > String x_field = row.getField("some_column").toString(); > //Do something with x_field > ... > tEnv.executeSql("INSERT INTO AnotherTable (column) VALUES > ('new_value')"); > } > } > > But this INSERT will be probably performance killer... > > Any suggestion how to do it in a smart way? > > Mike >
