Databricks want a piece of cake on the SQL-oneTrickPony Snowflake users. On their platform, they click a button and a compute instance appears in 2 seconds, and they can run sql.
they cannot orcherstrate jobs like workflows can. they cannot version their ML models like MlFlow can, they cannot version the sql code they love into github, no integration. the data they hold so dear is not in a bucket with a open-source table format that a lot of tools can query, is locked into the platform with a proprietary format that only a warehouse can query.
Databricks has a bright future because they build great products and make it EASY to use everything. But, I am troubled about certain things that are happening that are not open source or developer-friendly.
Databricks want a piece of cake on the SQL-oneTrickPony Snowflake users. On their platform, they click a button and a compute instance appears in 2 seconds, and they can run sql.
they cannot orcherstrate jobs like workflows can. they cannot version their ML models like MlFlow can, they cannot version the sql code they love into github, no integration. the data they hold so dear is not in a bucket with a open-source table format that a lot of tools can query, is locked into the platform with a proprietary format that only a warehouse can query.
they are lost man.
Databricks has a bright future because they build great products and make it EASY to use everything. But, I am troubled about certain things that are happening that are not open source or developer-friendly.
> And, contrary to what you may think, I believe Serverless options are indeed a wise choice in many instances.
Till your manager figures out that there are cheaper kind of compute and starts bothering you. Then it will tell you to go with an even cheaper one.
Their platform is becoming every day more overcomplicated, with more tools and options which are often half baked (despite what their marketing says)