Kibana as Business Intelligence?


(Edgar Colon) #1

Hello, I hope this should be the right forum if not, sorry for the inconvenience.

I'm in research for a platform that fulfill my needs for a Business Intelligence and ETL.

What we do have normally:
MySql 5.5 databases which stores all transactional information about our systems in tables which are in a relational model.
PHP 5.4 and Apache 2.2, all inside Centos 6.7 (components to run web apps we develop)

Those are my questions:

  1. Is Kibana what I need to do the BI and ETL job?
    If so,
  2. Where do I define metrics and dimensions?
  3. Where is the ETL? Do I need one? Which one?
  4. Is information taken from transactional database or creates a new one to do the analytics job?

I really appreciate your help.


(Tanya Bragin) #2

My take:

Maybe. There are some major differences between Kibana and traditional BI. BI tools generally connect to multiple backends, Kibana does not, it only connects to Elasticsearch. BI tools are aimed at fairly non-technical users (think Excel) whereas Kibana was originally developed for DevOps and IT Ops (though it's becoming easier to use and more customizable). You'd have to play around with the UI and be the judge.

Kibana does not have a UI for adding data to Elasticsearch. You must do it using Elasticsearch ingest APIs or something like Logstash. In the future, this will change (see Add Data UI issues in Github). But for now, you must do data ingest outside of Kibana.

Kibana does not connect to datastores other than Elasticsearch. If your information is in a database, you'd need to use something like a JDBC input in Logstash to index this data in Elasticsearch, and then create an index pattern in Kibana to look at that data. Keep in mind, Elasticsearch does not support joins, so you should take care to flatten the data prior to ingesting it, to take full advantage of Kibana.


(system) #3