Hi, According to the last question I know there is data locality mechenism when running Spark on the standalone mode sharing the same datanode servers with the HDFS cluster, so what other advantages we can achieve when running Spark inside a YARN cluster? Regards, Xiaobo Gu
