You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@airflow.apache.org by "Joy Gao (JIRA)" <ji...@apache.org> on 2018/03/21 22:08:00 UTC
[jira] [Commented] (AIRFLOW-2075) Extract and create a data access
layer
[ https://issues.apache.org/jira/browse/AIRFLOW-2075?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16408661#comment-16408661 ]
Joy Gao commented on AIRFLOW-2075:
----------------------------------
Creating a data access layer is a bit daunting, given it will touch pretty much every core module in airflow, which was built with the assumption where sqlachemy is the de facto ORM.
The alternative is to map Spanner onto a SQLAlchemy dialect, which will benefit not only Airflow but all projects that use sqlalchemy.
> Extract and create a data access layer
> --------------------------------------
>
> Key: AIRFLOW-2075
> URL: https://issues.apache.org/jira/browse/AIRFLOW-2075
> Project: Apache Airflow
> Issue Type: Improvement
> Components: core
> Affects Versions: Airflow 2.0
> Reporter: Yang Pan
> Priority: Major
>
> Currently Airflow is tightly coupled with sqlalchemy.
> [https://github.com/search?p=2&q=%22from+sqlalchemy%22+repo%3Aapache%2Fincubator-airflow+path%3A%2Fairflow&type=Code&utf8=%E2%9C%93]
> While it's convenient for sqlalchemy compatible database, it makes non compatible ones, such as google cloud spanner, harder to integrate.
> It would be nice if the data access layer can be decoupled from the airflow core, and developers have option to use sqlalchemy or swap to talk to db directly.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)