On Mon, Aug 7, 2023 at 6:03 AM Jialiang Cai <jialiangca...@gmail.com> wrote: > > Hello friends of the Ambari community, I am just an ordinary contributor, a > nobody. However, I would like to share some thoughts with you all. > > Ever since HDP was acquired, there has been no significant big data > distribution other than the HDP stack. The packing code and big data > components of HDP have become closed source. While many old users are still > using the HDP2 and HDP3.0 big data distributions, they no longer have the > corresponding component source code. This is a frightening situation, as the > lack of access to the source code for debugging when a component fails means > we have lost control over the data platform!!
That is an unfortunate situation indeed, but I guess inevitable (given that even RedHat is now engaging in similar behaviour). But to your point -- unlike Linux distributions -- where one can always go to something like Debian, for bigdata it is Bigtop. > Thus, I've been hoping that the community could collaborate with the Apache > Bigtop community to release the first fully open-source big data distribution > since HDP. Why Bigtop? Because the Apache Bigtop community is very active > with many developers maintaining package compatibility across different > platform architectures. Because of this, the Bigtop stack is a great choice. > Devoted community PMC member, Wuzhiguo, has made immense efforts in > developing Ambari 2.8.0 based on the Bigtop stack. Although he initiated a > vote, sadly the process has stalled for reasons. Huge +1 to the above proposal! > The active contributors in the community, have been able to fix a few bugs on > top of 2.8.0 and added some new components, aligning it with Apache Bigtop > 3.3.0. We aim to release these fixes in Ambari 2.9, striving for the first > stable, fully open-source, and user-friendly big data distribution since HDP. > Here is the version issue for Apache Bigtop components: > > https://issues.apache.org/jira/browse/BIGTOP-3909 > > We've conducted extensive testing for Ambari 2.8.0, subsequent master > branches as well as some important PR. We have also run it in a production > environment. Except for recurrent hive 3 out of memory issues, all other > components run smoothly. The hive 3 OOM issue has been submitted by another > contributor to Apache Bigtop and will be resolved in the Apache Bigtop 3.2.1 > and 3.3.0 releases. > > Members of the Ambari community who are still active have been working hard > to achieve the release of a new version of Ambari based on the Hadoop 3 > ecosystem. It maybe the case that a 'call for volunteers' on d...@community.apache.org, dev@[individual projects that are part of the Bigtop] or even members@a.o would be a very useful thing. Especially with a preamble of losing important bits of technology to close-source implementations of HDP. > We are limited in resources and sincerely hope to receive support from other > community members and precious advice and support from the Ambari community > PMCs. Without PMC's participation, the community will stagnate, and no > new Ambari releases will be published, a terrible outcome that none of us > want to see. It would be a shame if Ambari were to retire again due to lack > of activity, or if Ambari became outdated and unable to keep up with other > big data community developments, causing it to lose users. > > Therefore, I sincerely hope that everyone can actively participate, discuss > the release of the new version of Ambari, and the needed content. Your > valuable opinions and feedback are greatly appreciated. The main content of > the next issue, based on the Ambari version of Hadoop 3, is here: > > https://issues.apache.org/jira/browse/AMBARI-25951 > > We hope that the PMCs will look it over and share their valuable suggestions. Thanks, Roman. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@ambari.apache.org For additional commands, e-mail: dev-h...@ambari.apache.org
