Why does Helix allow different way to modify cluster config?

[Henry Saputra <he...@gmail.com>]

Hi Guys,

I have testing and play around with Helix and I have one question:

As shown in the tutorial page [1] why does Helix allow two different
ways, one is via Java accessor methods in the class structure and the
other via HelixAdmin, to modify the cluster configuration through the
API? The page did not mention why we should choose one over the other.
I think it adds confusion instead of maybe the intention of flexibility.

Would this be addressed with the new admin API work?

Thanks,

- Henry

[1] http://helix.apache.org/trunk-docs/Tutorial.html

RE: Why does Helix allow different way to modify cluster config?

[Kanak Biscuitwala <ka...@hotmail.com>]

Hi Henry,

Yes, we are planning to come up with a totally new administrator interface that more completely covers use cases, and then will deprecate HelixAdmin. The code currently in trunk and 0.7 is very unstable and was just a result of experiments and trying to learn what exactly is necessary for such an API.

If you're planning to put Helix in production, I would go with 0.6.3 for now.

Kanak

----------------------------------------
> Date: Fri, 21 Mar 2014 15:42:53 -0700
> Subject: Why does Helix allow different way to modify cluster config?
> From: henry.saputra@gmail.com
> To: dev@helix.apache.org
>
> Hi Guys,
>
> I have testing and play around with Helix and I have one question:
>
> As shown in the tutorial page [1] why does Helix allow two different
> ways, one is via Java accessor methods in the class structure and the
> other via HelixAdmin, to modify the cluster configuration through the
> API? The page did not mention why we should choose one over the other.
> I think it adds confusion instead of maybe the intention of flexibility.
>
> Would this be addressed with the new admin API work?
>
> Thanks,
>
> - Henry
>
> [1] http://helix.apache.org/trunk-docs/Tutorial.html