danny0405 commented on code in PR #11555: URL: https://github.com/apache/hudi/pull/11555#discussion_r1678672913
########## rfc/rfc-79/rfc-79.md: ########## @@ -0,0 +1,108 @@ +<!-- + 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-79: Improving reliability of concurrent table service executions and rollbacks + +## Proposers + +- @kbuci +- @suryaprasanna +- @nsivabalan + +## Approvers + +## Status + +JIRA: HUDI-7946 + + +## Abstract +In order to improve latency/throughput of writes into a HUDI dataset, HUDI does not require that table service operations (such as clustering and compaction) be serially and sequentially performed before/after an ingestion write. Instead, by enabling HUDI multiwriter and async table service execution, a user can orchesterate seperate writers to potentially execute table service plans concurrently to an ingestion writers. This setup though may face reliability issues for clustering and compaction, as failed executions and rollbacks can cause delays in table service executions and prevent cluster/compaction/clean operations from being scheduled. This RFC proposes to address these limitations by using HUDI's heartbeating and transaction manager to update the behavior for clustering, compaction, and rollback of failed writes. With these changes users can build an orchestration platform for executing each table service independently without needing to make complicated/expensive changes t o prevent multiple job/threads for targeting the same table service plan. Review Comment: > In order to improve latency/throughput of writes into a HUDI dataset, HUDI does not require that table service operations (such as clustering and compaction) be serially and sequentially performed before/after an ingestion write This is not true since 1.x, by using the `TrueTime` API, the compaction plan can be scheduled in any separate job instead of in the intestion writer. -- 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]
