We use redash. This allows forking of queries, charting, and some light dashboarding. Also it supports full text searching of queries and descriptions. It's also stand alone and easy to launch on ec2.
We directly use an ipython notebook server shared for the whole company. A lot more flexible but require a big level of trust since it allow total system control
That looks great! I very much prefer the canonical SQL for querying data in a visual way, rather than a "generic" custom querying language. Does it send the query directly to the db? Is there any transformation? Does it mean it only works with sql databases?
Yes, Redash sends the query as is to the database. It currently supports PostgreSQL (& Redshift), MySQL, BigQuery, MongoDB, Graphite and Python scripts. Also someone currently working on ElasticSearch support.
The MonogDB & Graphite support is crude, but exists.
I'm the project maintainer, feel free to ask if you have any questions.
This looks great! I have a number of adhoc queries I keep kicking around. Some of which them are timeseries based and I've taken the time to pump those into graphite, but others are just interesting to see in their own right sometimes.
I have the exact same feeling and question. The copy "Get data yourself. No need to bug engineers anymore." confuses me further. Surely someone who can write SQL queries can setup PgCommander or something similar?
This looks pretty awesome! It is also worth mentioning Django SQL Explorer (https://github.com/epantry/django-sql-explorer), which has all of the presented features, as well as handy schema definition by the side. It also allows for parameters in the queries, which is very useful (e.g. leaving the timeframe customisable before running a query). With Django 1.8 now being easier to install/deploy to wherever, this is a great plugin to allow more people data access!
I thought that'd be an abstraction layer so people who are not that familiar with SQL but need some basic information can easily get it but this doesn't seem any different than just giving someone access to a view with a regular GUI client or http://pgcli.com/.
We use blazer[1] which is open source and lets you share queries just like this video demonstrates. It also has other nice features like composing results into charts and allowing interpolated values for generating queries with variable inputs. The demo[2] has examples for the features I mentioned.
How did you guys drop down to a read-only connection on Heroku Postgres? We directly connect DB clients like PGAdmin to a reporting DB on Heroku, but Heroku only seems to support credentials with full access. I don't want an analyst to "DROP TABLES" by mistake.
https://github.com/EverythingMe/redash