https://community.cloudera.com/t5/Support-Questions/kudu-is-slower-than-parquet/m-p/56504#M14061 <P>We'd expect Kudu to be slower than Parquet on a pure read benchmark, but not 10x slower - that may be a configuration problem. We've published results on the Cloudera blog before that demonstrate this:&nbsp;<A href="http://blog.cloudera.com/blog/2017/02/performance-comparing-of-different-file-formats-and-storage-engines-in-hadoop-file-system/&nbsp;" target="_blank">http://blog.cloudera.com/blog/2017/02/performance-comparing-of-different-file-formats-and-storage-engines-in-hadoop-file-system/&nbsp;</A></P><P>&nbsp;</P><P>Parquet is a read-only storage format while Kudu supports row-level updates so&nbsp;they make different trade-offs. I think we have headroom to significantly improve the performance of both table formats in Impala over time.</P><P>E.g. in Impala 2.9/CDH5.12&nbsp;<A href="http://issues.cloudera.org/browse/IMPALA-5347" target="_blank">IMPALA-5347</A>&nbsp;and&nbsp;IMPALA-5304 improve pure Parquet scan performance by 50%+ on some workloads, and I think there are probably similar opportunities for Kudu.</P><P>&nbsp;</P> Mon, 26 Jun 2017 15:41:17 GMT Tim Armstrong 2017-06-26T15:41:17Z