+1 2025年9月24日(水) 7:19 Mich Talebzadeh <[email protected]>:
> well. I see Nimrod has a valid point here. one can hide “Spark Connect vs > classic Spark SQL (HiveServer2/Beeline/Thrift) from users by putting a thin > abstraction in front. > Once that abstraction layer is there, then the hand over becomes > transparent. Your app talks to a Spark Connect endpoint, the Spark driver > runs in the cluster. Works with DataFrame API and SQL. > In essence you design a small library (or service) that exposes one API to > users and chooses the backend under the bonnet. > > HTH > > Dr Mich Talebzadeh, > Architect | Data Science | Financial Crime | Forensic Analysis | GDPR > > view my Linkedin profile > <https://www.linkedin.com/in/mich-talebzadeh-ph-d-5205b2/> > > > > > > On Tue, 23 Sept 2025 at 21:40, Nimrod Ofek <[email protected]> wrote: > >> Hi, >> >> That's the thing - I don't expect the users to know if they are >> connecting to Spark or Spark connect. >> That means I would expect the driver to support both Spark Connect - and >> the current Hive/ Beeline /Thrift server. >> Maybe even some regular Spark API (even a simple "jar" that just runs the >> query and returns the results somehow, haven't really thought about it in >> depth). >> >> The point I'm trying to make is that the user that wants to run a SQL >> command using JDBC - doesn't care if it's Spark or Spark connect or >> whatever - it's like a Database for the user... >> >> Regards, >> Nimrod >> >> >> On Mon, Sep 22, 2025 at 5:24 PM Cheng Pan <[email protected]> wrote: >> >>> Hi Nimrod, >>> >>> I'm not sure I get your question. Maybe the name should be 'JDBC >>> Driver for Spark Connect Server'? >>> >>> From the user's perspective, they simply use a JDBC driver to connect >>> to the Connect Server to run SQL >>> and retrieve the results, without having to worry about whether Spark >>> is running in classic or Connect mode. >>> >>> Thanks, >>> Cheng Pan >>> >>> On Mon, Sep 22, 2025 at 10:17 PM Nimrod Ofek <[email protected]> >>> wrote: >>> > >>> > I'll raise an issue with this- I don't think the user that uses jdbc >>> to Spark should know if he is working with Spark connect or regular >>> Spark.... >>> > The jdbc driver should know how to work with connect with fallback >>> maybe, but the user doesn't care if he is getting Spark connect or not... >>> > >>> > Regards, >>> > Nimrod >>> > >>> > בתאריך יום ב׳, 22 בספט׳ 2025, 16:04, מאת 杨杰 <[email protected]>: >>> >> >>> >> Hi Spark devs, >>> >> >>> >> I would like to start a vote on the SPIP: JDBC Driver for Spark >>> Connect >>> >> >>> >> Discussion thread: >>> >> https://lists.apache.org/thread/rx5pqh01c86slpqv9161hqwgm5lwxxzq >>> >> SPIP: >>> >> >>> https://docs.google.com/document/d/1Ahk4C16o1Jj1TbLg5ylzgHjvu2Ic2zTrcMuvLjqSoAQ/edit?tab=t.0#heading=h.1gf0bimgty0t >>> >> JIRA: https://issues.apache.org/jira/browse/SPARK-53484 >>> >> >>> >> Please vote on the SPIP for the next 72 hours: >>> >> >>> >> [ ] +1: Accept the proposal as an official SPIP >>> >> [ ] +0 >>> >> [ ] -1: I don’t think this is a good idea because >>> >>
