Import Substitute Hadoop Distribution
Recently, looking at the registry of domestic software, I found the Russian assembly Hadoop, to which I was pretty surprised.
Moreover, the registration was already in 2017. ( link for those interested ) and since then there is not a single line on this in Habré
So, welcome - Arenadata Hadoop (ADH)!
First impressions:
I went to the site and ... The last time I experienced such a combination of perplexity and perplexity was in 2014, when I met Belarusian oysters in a store.
Judge for yourself:
here is the site Arenadata
here is the Hortonworks site
')
Primary colors are green and gray.
Both distributions use Ambari to manage the cluster, only ADP has the inscription Arenadata, and Horton does not:
After that I went to the Roadmap section, see the versions of the components and the first reaction was “why did you need to transpose the table?”:
Option from Horton:
And the version of Arenadata:
After that, it became a bit sad and decided that the “import substitution” ended with a re-glued label, especially noting Greenplum as Arenadata DB and Arenadata Grid in the product list, but then carefully looked at the package versions and realized that the basis was taken from Hortonworks, and its content , and it pleases.
A small digression: yes, I know that Hortonworks and Cloudera have merged, but the company will release distributions under the same brand names for another 3 years, so I don’t call Hortonworks Clouder, although I could.
Of the benefits - there is Russian-language documentation and support (newbies and those who worked with support in the form of Hindus in America - should appreciate), however, support by subscription, and I’m not a forum where you can ask questions about the painful Cloudera / Horton found.
And now there are some oddities: there are 2 versions on the site : Enterprise and Platform, and Enterprise is pretty incomprehensible - without Spark, Zeppelin and Ranger, and only one version is available for download and that one without the right to choose.
However, judging by the presence of Ranger, the most complete build of the Arenadata Hadoop Platform is available.
But it is, little things.
Only two things are interesting: when ADP 2.0 comes out of release with Hadoop 3 under the hood, give Yarn-on-GPU that can compete with releases from Cloudera and the quality of support — can it dig up and solve a low-level bug, or will the Apache Foundation ticket be launched? will need to wait for an answer.
PS: I did not download and locally install the distribution for one simple reason: I was used to working with multi-mode clusters, and after a local installation (read, no thousands of cores and terabytes of RAM), the impression would be pretty spoiled.
Moreover, the registration was already in 2017. ( link for those interested ) and since then there is not a single line on this in Habré
So, welcome - Arenadata Hadoop (ADH)!
First impressions:
I went to the site and ... The last time I experienced such a combination of perplexity and perplexity was in 2014, when I met Belarusian oysters in a store.
Judge for yourself:
here is the site Arenadata
here is the Hortonworks site
')
Primary colors are green and gray.
Both distributions use Ambari to manage the cluster, only ADP has the inscription Arenadata, and Horton does not:
After that I went to the Roadmap section, see the versions of the components and the first reaction was “why did you need to transpose the table?”:
Option from Horton:
And the version of Arenadata:
After that, it became a bit sad and decided that the “import substitution” ended with a re-glued label, especially noting Greenplum as Arenadata DB and Arenadata Grid in the product list, but then carefully looked at the package versions and realized that the basis was taken from Hortonworks, and its content , and it pleases.
A small digression: yes, I know that Hortonworks and Cloudera have merged, but the company will release distributions under the same brand names for another 3 years, so I don’t call Hortonworks Clouder, although I could.
Of the benefits - there is Russian-language documentation and support (newbies and those who worked with support in the form of Hindus in America - should appreciate), however, support by subscription, and I’m not a forum where you can ask questions about the painful Cloudera / Horton found.
And now there are some oddities: there are 2 versions on the site : Enterprise and Platform, and Enterprise is pretty incomprehensible - without Spark, Zeppelin and Ranger, and only one version is available for download and that one without the right to choose.
However, judging by the presence of Ranger, the most complete build of the Arenadata Hadoop Platform is available.
But it is, little things.
Only two things are interesting: when ADP 2.0 comes out of release with Hadoop 3 under the hood, give Yarn-on-GPU that can compete with releases from Cloudera and the quality of support — can it dig up and solve a low-level bug, or will the Apache Foundation ticket be launched? will need to wait for an answer.
PS: I did not download and locally install the distribution for one simple reason: I was used to working with multi-mode clusters, and after a local installation (read, no thousands of cores and terabytes of RAM), the impression would be pretty spoiled.
Source: https://habr.com/ru/post/454416/
All Articles