Plans for interface to reporting Tools

Are there any plans in the Druid roadmap to provide a ODBC or JDBC driver so that tools like Tableau can be connected to Druid?

Ted,

SQL4D has a jdbc driver implemented:

I'm also trying to connect with Tableau to see about their javascript
data connector API that I heard you guys actually leveraged for
connecting to Presto. If you have any information that you can share
about that, I'd love to hear it :).

--Eric

Ah ok, I was aware of Sql4D, but didn’t realize it was functional as a jdbc driver based on what I read in this thread:
https://groups.google.com/d/msg/druid-user/qcYXGwXlB-k/ZoOdKPZuonkJ

I’m discussing with some of my colleagues about how they have the presto driver working. I’ll update this thread if anything interesting that I can share comes out of it.

-Ted

Hi Eric,

How’s the Tableau thing going? I’ve just tried it out,the vizql model is really cool! Bad things is it’s not for free :frowning:

在 2015年6月9日星期二 UTC+8上午4:20:37,Eric Tschetter写道:

Hi Weinan, there are folks working on a more complex open source analytics front end for Druid. There are also open source options available now, for example, Gian recently did a tutorial for Grafana on Druid:

Hi Fangjin,

I’ve heard about the Druid plugin for Grafana and used it in our team, it is easy to use, has a cool UI, but suitable for monitor/dashboard scenario which is much less than Druid’s capability. I think may be a data exploration and visualization tool like Tableau will liberate the power of druid.

在 2015年7月16日星期四 UTC+8上午7:55:48,Fangjin Yang写道:

Hi Weinan, absolutely agree. There’s no great open source exploration tool for Druid just yet, but I hope that’s something that changes in the near future :slight_smile:

Hi there,

I’m trying to decide whether Druid is right for our organization and the lack of SQL support is a major stumbling block for integrating with our other tools. I’m wondering if anything has changed in this regard since this thread was last updated. The SQL4D github repo hasn’t been touched in 11 months, and it is unclear how complete/stable the Spark SQL integration is (only 1 contributor). I have seen, but haven’t investigated Caravel and Pivot as analytics tools, but it would be better if we could use the tools we have already. What are people doing today to analyze the data they have in Druid?

Thanks,

–Ben

Hey Ben,

PlyQL is currently the most well maintained SQL layer for Druid: http://github.com/implydata/plyql. It does not come with an ODBC/JDBC connector though, so while you can issue SQL queries, you would have to do a bit of leg work to integrate it with something that expects a connector.

That being said – I definitely would recommend looking into Pivot and Caravel if you have a chance. They are both well integrated with Druid and provide a nice interactive query experience.

Hi Ben,

You might be interested to check out: http://imply.io/post/2016/05/04/programmatic-plyql.html and see how far what will take you.

Vadim

Thank you! I actually saw the commit today that added JDBC to PlyQL and my team was actually discussing it right now. Very exciting!

–Ben

As it says in the bottom of the post: any and all feedback about queries that fail to deliver correct results for one reason or another are welcome.

Thank you!

I realize this is an experimental feature (the mysql gateway) so I’m curious as to your opinion of its stability and performance. I’m less worried about completeness because most of our use cases are simple selects with some aggregation and a simple where clause which it sounds like it should be supported.

Hi Ben, we have not done enough testing to make a statement about performance and stability at this point.

PlyQL itself is quite widely used and seems to be pretty stable. The MySQL gateway is less well known.

All JDBC/ODBC drivers have so sort of handshake that they use to get metadata about the database.

It is one thing to answer a question like: “SELECT page, COUNT(*) FROM wikipedia GROUP BY page LIMIT 20” (that is a topN) it is quite another beast to answer:

“SHOW FULL COLUMN IN wikipedia” or “SELECT @@global.max_allowed_packet, @@default_server_charset

So far the focus has been on making these handshake/metadata queries work. I would be very interested if your driver of choice did some handshake that PlyQL could not handle.

I think the stability of the whole gateway should be better known in a month or two.

Hi there,

I have been trying to play with this functionality with mixed results. Connecting from mysql’s CLI has worked well, but connecting from one of our BI tools (Metric Insights) is failing. The output from the mysql gateway is:

[5] Got SQL: /* mysql-connector-java-5.1.36 ( Revision: 4fc1f969f740409a4e03750316df2c0e429f3dc8 ) */SELECT @@session.auto_increment_increment, @@character_set_client, @@character_set_connection, @@character_set_results, @@character_set_server, @@init_connect, @@interactive_timeout, @@license, @@lower_case_table_names, @@max_allowed_packet, @@net_buffer_length, @@net_write_timeout, @@query_cache_size, @@query_cache_type, @@sql_mode, @@system_time_zone, @@time_zone, @@tx_isolation, @@wait_timeout

[5] Got SQL: SELECT @@session.autocommit

[5] Got SQL: SET NAMES utf8mb4

[5] Got SQL: SET character_set_results = NULL

[5] Got SQL: SET autocommit=1

[5] Error: Connection lost: The server closed the connection.

And I get an error on the client:

Packet for query is too large (45 > 1). You can change this value on the server by setting the max_allowed_packet’ variable.

Any ideas?

Thanks,

–Ben

One further detail after further exploration; it appears to be running a “select now();” query when it gets the error, although it isn’t output on the plyql console.

–Ben

Great feedback, will look into it. Could you please post an issue here: https://github.com/implydata/plyql/issues so we can keep track of it.

Hi Ben,

I am looking for a similar solution of connecting druid to tableau or BI tool that expects JDBC/ODBC connectors. Could you please provide any pointers or details on how you were able to address this issue. I am also interested if you were able to use JDBC connector provided in PlyQL as part of recent commits.

I have been trying Sparkline, but ran into few issues while querying druid from tableau. Any help would be appreciated.

Thank you.

-Manasa

Hi Manasa,

I was unable to get Tableau to work through PlyQL. The inspection Tableau performs for understanding the schema does not work. My other BI tools are working as they are using simpler SQL statements. We have made the decision to move away from Tableau for unrelated reasons so it hasn’t been a big priority for us. Sorry that isn’t very helpful.

–Ben

Hi Ben,

Thank you though. Would you be able to share which BI tool you were able to get druid working. I could use caravel, haven’t tried Grafana, but would like to check out other options. Thank you.

-Manasa