wombatu-kun commented on code in PR #11559: URL: https://github.com/apache/hudi/pull/11559#discussion_r1699402996
########## rfc/rfc-80/rfc-80.md: ########## @@ -0,0 +1,187 @@ +<!-- + Licensed to the Apache Software Foundation (ASF) under one or more + contributor license agreements. See the NOTICE file distributed with + this work for additional information regarding copyright ownership. + The ASF licenses this file to You under the Apache License, Version 2.0 + (the "License"); you may not use this file except in compliance with + the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + + Unless required by applicable law or agreed to in writing, software + distributed under the License is distributed on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + See the License for the specific language governing permissions and + limitations under the License. +--> +# RFC-80: Support column families for wide tables + +## Proposers +- @xiarixiaoyao +- @wombatu-kun + +## Approvers + - @vinothchandar + - @danny0405 + - @beyond1920 + +## Status + +JIRA: https://issues.apache.org/jira/browse/HUDI-7947 + +## Abstract + +In streaming processing, there are often scenarios where the table is widened. The current mainstream real-time wide table concatenation is completed through Flink's multi-layer join; +Flink's join will cache a large amount of data in the state backend. As the data set increases, the pressure on the Flink task state backend will gradually increase, and may even become unavailable. +In multi-layer join scenarios, this problem is more obvious. +1.x also supports partial updates being encoded in logfiles. That should be able to handle this scenario. But even with partial-update, the column families will reduce write amplification on compaction. + +So, main gains of clustering columns for wide tables are: +Write performance: +- Writing is similar to ordinary bucket writing, but it involves splitting column clusters and sorting. Therefore, the writing performance of full data is 10% lower than that of native bucket writing. +- However, if only some columns are updated among a large number of columns, the writing efficiency is much faster than that of non-column clustered tables. (Note: 1.0 introduces a partial update functionality, that can also avoid such writing costs, by writing only the changed columns) + +Read performance: +Since the data is already sorted when it is written, the SortMerge method can be used directly to merge the data; compared with the native bucket data reading performance is improved a lot, and the memory consumption is reduced significantly. Review Comment: good idea, but may be we can do it later as a follow-up in separate task? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
