Hi all.
Apache Doris (Incubating) has been in development for over 3 years since it
entered the Apache incubator in September 2018.
During our time in the Incubator, we have learned a lot about open source
software governance and the Apache Foundation's philosophy on managing,
maintaining, and developing open source community.
I am pleased to conduct a discussion here on matters related to Doris'
graduation from the incubator.
The following is a brief overview of the progress of the Doris project and
community since entering the incubator.
1. Community
- 8 new PPMC members were added, from five different companies, bringing
the total number of PPMC members to 21.
- 16 new Committers were added (including the new PPMC members), bringing
the total number of Committers to 29.
- The number of Contributors is now 245 and growing.
- The dev@doris mailing list currently has 289 subscribers, and all major
project discussions are happening in the dev@doris.
2. Project
- 7 releases by 6 release managers. All compliance issues have been
resolved.
- Doris official website[1] is compliant with Apache Foundation
requirements[2].
- Project maturity model is detailed in [3].
- We have also created a pre-graduation Check List[4] following the Apache
official website guidelines and are currently moving forward.
Some things that I think need further discussion.
1. Branding issue
Doris had some branding issues during the incubating[5][6]. I think these
issues have been resolved and would appreciate the mentors' assistance in
checking them out.
2. Relationship with Apache Impala
Doris' current query engine is forked from Impala version v0.7 and has been
updated with Impala in the early stages of the project (hands-on code merge).
In the Doris Proposal [7], we expressed our desire to decouple the Doris
storage engine from the query engine through a layer of abstract API. So that
the query layer could fully follow Impala's evolution.
However, as the project progressed, we found it difficult to achieve this
goal, mainly for the following reasons.
1. Impala focus on providing query acceleration on the Hadoop ecosystem,
while Doris focus on a standalone data warehouse or DBMS system. The difference
of targets means that many features do not meet the needs of both projects.
2. Doris has completely refactored the metadata and has its own storage
engine. This required a deep integration of the query layer with the metadata
and the storage engine. We also made a lot of architectural changes for this.
3. Doris has now developed a new vectorized execution engine that rewrites
most of the execution node in query layer, which is different from Impala.
For these reasons, it is difficult for us to continue to follow the Impala
community with code level iterations and integrations.
Of course, all code references to Impala are fully compliant with the
Apache License, and we explicitly declare the relationship between Doris and
Impala in the NOTICE file[8] and the README file[9] and thank the Impala
community for their help.
Since this aspect deviates from the direction in the original Proposal, I
think it needs to be listed separately for necessary discussion.
If there are any other aspects that need to be discussed before graduation,
please discuss.
[1] http://doris.incubator.apache.org/
[2] https://whimsy.apache.org/pods/project/doris
[3]
https://cwiki.apache.org/confluence/display/DORIS/Maturity+Assessment+for+Doris
[4] https://cwiki.apache.org/confluence/display/DORIS/Graduation+Check+List
[5] https://lists.apache.org/thread/k8klobjzffobbx161nj6plc0mfl8gc7z
[6] https://lists.apache.org/thread/t6dwq77x0jjm63kpkzlq51sfbowd92p9
[7] https://cwiki.apache.org/confluence/display/incubator/DorisProposal
[8] https://github.com/apache/incubator-doris/blob/master/NOTICE.txt
[9] https://github.com/apache/incubator-doris/blob/master/README.md
--
此致！Best Regards
陈明雨 Mingyu Chen
Email:
chenmin...@apache.org
