[ https://issues.apache.org/jira/browse/SPARK-13125?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15129860#comment-15129860 ]
zhengcanbin commented on SPARK-13125: ------------------------------------- Yes, you are right, but shuffle will increase net burden, and number of partitions is limited by total number of disk. In strictly real-time scenarios, one topic partition corresponds to multiple rdd partitions is important for increasing parallelism > makes the ratio of KafkaRDD partition to kafka topic partition configurable. > ----------------------------------------------------------------------------- > > Key: SPARK-13125 > URL: https://issues.apache.org/jira/browse/SPARK-13125 > Project: Spark > Issue Type: Improvement > Components: Input/Output > Affects Versions: 1.6.1 > Reporter: zhengcanbin > Labels: features > Original Estimate: 96h > Remaining Estimate: 96h > > Now each given Kafka topic/partition corresponds to an RDD partition, in some > case it's quite necessary to make this configurable, namely a ratio > configuration of RDDPartition/kafkaTopicPartition is needed. -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org