[
https://issues.apache.org/jira/browse/STORM-3497?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16930012#comment-16930012
]
Stig Rohde Døssing commented on STORM-3497:
-------------------------------------------
[~liketic] Sorry for not getting back to you sooner.
Not really sure what you mean by full integration? Is there some part of the
connection logic you can't implement with a spout and bolt?
I think you are likely right that a connector for your service will be useful
to your customers. My concern is that putting it in the Storm repository will
make it hard to maintain. Keep in mind that if you put the connector in the
Storm repository, any updates need to go through review by a Storm committer.
If the reviewer aren't familiar with your service, and has a testing
environment, any review is likely going to be slow and inaccurate. Worse,
changes may end up not being reviewed, and thus not getting merged.
We've seen this happen with other connectors, where it would probably have been
better if the connector were maintained by the company offering the service,
instead of being part of Storm's repo. If you maintain the connector in your
own repo, you have more flexibility to decide what you think are appropriate
changes to the connector, and can do releases independently of Storm's release
cadence.
If someone currently maintaining Storm were familiar with your service and
willing/able to review changes, it would be fine to add it to Storm's repo. I
just don't want us to add the connector and have it end up not maintained. This
is why I think asking on the mailing list makes sense. If no one with commit
rights to Storm knows enough about your service to maintain the connector, I
think it should be maintained separately from Storm, until either someone on
the committer list starts using your service, or you/one of your coworkers get
commit rights to Storm. Otherwise we end up it a situation where the people
with merge rights can't properly evaluate pull requests to your connector,
because they aren't familiar enough with the service to say whether a change is
good or bad.
> Add connector for Alibaba Log Service
> -------------------------------------
>
> Key: STORM-3497
> URL: https://issues.apache.org/jira/browse/STORM-3497
> Project: Apache Storm
> Issue Type: New Feature
> Components: storm-core
> Reporter: Ke Li
> Priority: Major
>
> Alibaba Log Service is a big data service which has been widely used in
> Alibaba Group and thousands of customers of Alibaba Cloud. The core storage
> engine of Log Service is named Loghub which is a large scale distributed
> storage system, which provides producer and consumer to push and pull data
> like AWS Kinesis or Azure Eventhubs does.
> Log Service provides a complete solution to help user collect data from both
> on premise and cloud data sources. More than 10 PB data is sent to and
> consumed from Loghub every day. And hundreds of thousands of users of Alibaba
> are working with Log Service, and Storm is an important component for
> building their big data system.
> We'd like to contribute a connector to Storm. Happy to hear any comments.
> Thank you in advance.
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
