You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@zeppelin.apache.org by sourav-mazumder <gi...@git.apache.org> on 2016/03/16 18:29:08 UTC
[GitHub] incubator-zeppelin pull request: [ZEPPELIN-513] Dedicated interpre...
Github user sourav-mazumder commented on the pull request:
https://github.com/apache/incubator-zeppelin/pull/703#issuecomment-197444164
Hi,
I was just trying some of these. Looks like with this solution there would be queue built up for running multiple paragraphs in parallel for Spark. And that essentially causes huge slowness when multiple paragraphs are running in parallel. It is a big road block for multi-tenancy.
My suggestion is to create multiple processes for Spark Interpreter but limit it/link it to user instead of notebook. So for every user there should be a separate interpreter process instead of for every notebook. As we are also having PR on user authentication and authentication in place achieving this should not be a problem.
Regards,
Sourav
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
Re: [GitHub] incubator-zeppelin pull request: [ZEPPELIN-513] Dedicated interpre...
Posted by "Amos B. Elberg" <am...@gmail.com>.
This links to 703, which creates per-notebook interpreters and I'm now starting to see problems with.
> On Mar 16, 2016, at 1:29 PM, sourav-mazumder <gi...@git.apache.org> wrote:
>
> Github user sourav-mazumder commented on the pull request:
>
> https://github.com/apache/incubator-zeppelin/pull/703#issuecomment-197444164
>
> Hi,
>
> I was just trying some of these. Looks like with this solution there would be queue built up for running multiple paragraphs in parallel for Spark. And that essentially causes huge slowness when multiple paragraphs are running in parallel. It is a big road block for multi-tenancy.
>
> My suggestion is to create multiple processes for Spark Interpreter but limit it/link it to user instead of notebook. So for every user there should be a separate interpreter process instead of for every notebook. As we are also having PR on user authentication and authentication in place achieving this should not be a problem.
>
> Regards,
> Sourav
>
>
> ---
> If your project is set up for it, you can reply to this email and have your
> reply appear on GitHub as well. If your project does not have this feature
> enabled and wishes so, or if the feature is enabled but not working, please
> contact infrastructure at infrastructure@apache.org or file a JIRA ticket
> with INFRA.
> ---