Feature about taking more than one hive source as kylin input in kylin-1.x

[yu feng <ol...@gmail.com>]

Hi all:
    I have been submit a patch on kylin-1.0(KYLIN-1172), it can take more
hive as kylin input. even those hive based on different hadoop cluster.
However, After I consult with @Luke Han and @Shi, Shaofeng, they do not
plan to add new big fearures to kylin-1.x which is in a stable state.

    I think maybe we can create a new branch for this feature, and offer an
option for those who really need it. Instead of deploying many kylin
environment for different hive source.

    What is more, I am being familiar with kylin-2.x which is more
extensible in architecture, and I will add one input named "other hives"
which can realize the same feature. and I think it will easier to add this
feature to kylin-2.x.

    I want to get some advice from kylin community and what is next for me
to create a new branch. Thanks for any suggestion.

Re: Feature about taking more than one hive source as kylin input in kylin-1.x

["Shi, Shaofeng" <sh...@ebay.com>]

Maintain additional branch is time-consuming, you have to periodically
merge it with the main branch; I think for 1.x the patch would be good,
user can pick up it and apply if he need.

On 1/2/16, 12:18 PM, "Luke Han" <lu...@gmail.com> wrote:

>I think a new branch is good idea for people who would like to have this
>feature, but it also need continue maintain like merge bug fixes from
>1.x-staging and future release.
>
>For new branch, you could create branch in your cloned code repo first and
>asking community to try, and then submit to main git repo later when you
>have permission to push code.
>
>Since this patch requires a lot of changes on 1.x, and actually such
>changes has been made in 2.x say plug-able architecture. There's possible
>way to extend to any source rather than Hive, like SparkSQL or other SQL
>on
>Hadoop. And this feature brings another great opportunity to pulling data
>from different cluster.
>
>Looking forward for your new patch to 2.x ;-)
>
>Thanks again.
>
>Luke
>
>
>
>
>Best Regards!
>---------------------
>
>Luke Han
>
>On Sat, Jan 2, 2016 at 11:52 AM, yu feng <ol...@gmail.com> wrote:
>
>> Hi all:
>>     I have been submit a patch on kylin-1.0(KYLIN-1172), it can take
>>more
>> hive as kylin input. even those hive based on different hadoop cluster.
>> However, After I consult with @Luke Han and @Shi, Shaofeng, they do not
>> plan to add new big fearures to kylin-1.x which is in a stable state.
>>
>>     I think maybe we can create a new branch for this feature, and
>>offer an
>> option for those who really need it. Instead of deploying many kylin
>> environment for different hive source.
>>
>>     What is more, I am being familiar with kylin-2.x which is more
>> extensible in architecture, and I will add one input named "other hives"
>> which can realize the same feature. and I think it will easier to add
>>this
>> feature to kylin-2.x.
>>
>>     I want to get some advice from kylin community and what is next for
>>me
>> to create a new branch. Thanks for any suggestion.
>>

Re: Feature about taking more than one hive source as kylin input in kylin-1.x

[Luke Han <lu...@gmail.com>]

I think a new branch is good idea for people who would like to have this
feature, but it also need continue maintain like merge bug fixes from
1.x-staging and future release.

For new branch, you could create branch in your cloned code repo first and
asking community to try, and then submit to main git repo later when you
have permission to push code.

Since this patch requires a lot of changes on 1.x, and actually such
changes has been made in 2.x say plug-able architecture. There's possible
way to extend to any source rather than Hive, like SparkSQL or other SQL on
Hadoop. And this feature brings another great opportunity to pulling data
from different cluster.

Looking forward for your new patch to 2.x ;-)

Thanks again.

Luke




Best Regards!
---------------------

Luke Han

On Sat, Jan 2, 2016 at 11:52 AM, yu feng <ol...@gmail.com> wrote:

> Hi all:
>     I have been submit a patch on kylin-1.0(KYLIN-1172), it can take more
> hive as kylin input. even those hive based on different hadoop cluster.
> However, After I consult with @Luke Han and @Shi, Shaofeng, they do not
> plan to add new big fearures to kylin-1.x which is in a stable state.
>
>     I think maybe we can create a new branch for this feature, and offer an
> option for those who really need it. Instead of deploying many kylin
> environment for different hive source.
>
>     What is more, I am being familiar with kylin-2.x which is more
> extensible in architecture, and I will add one input named "other hives"
> which can realize the same feature. and I think it will easier to add this
> feature to kylin-2.x.
>
>     I want to get some advice from kylin community and what is next for me
> to create a new branch. Thanks for any suggestion.
>