Bigger picture

- Central repository for database audit logs
- Listener and alert logs to be parsed and stored in the central repository
- Performance metrics (AWR) for troubleshooting and capacity planning

 

- Possibility of Real-time analytics, Offline analytics and visualization
- Reusable open source solution

Architecture

Elastic Stack

Elasticsearch

- distributed, open source search and analytics engine

Kibana

 - open source data visualisation platform

 - histogram, geomaps, line and pie charts...

Kibana plugin - goals

- create a new type of visualisation

- possibility of choosing begin and end date

- metric name

- database id

Developing Kibana plugin

- each plugin is npm module

- package.json provides list of all dependencies

- npm install command installs the dependencies

 Timelion plugin's package.json

Developing Kibana plugin

 -  Yeoman generator provides basic structure of the plugin

mkdir my-new-plugin
cd my-new-plugin
yo kibana-plugin

- Generating plugin with yeoman

Developing Kibana plugin

 - adding new html files

 - new js files - controllers

 - creating new routes for getting the data from Elasticsearch

Next steps

Developing Kibana plugin

AngularJS

- powerful JavaScript framework

- extends HTML with ng-directives

- provides data binding

- controllers

Developing Kibana plugin

Kibana plugin - during the work...

Kibana plugin - during the work...

How to install the plugin

1) switch to Kibana plugin directory

            cd /opt/kibana/installedPlugins

 

2) clone the repository ( install the plugin )

            git clone https://github.com/MsSquirrel/OracleLogs.git

 

3) install dependencies mentioned in package.json

             npm install

 

4) restart Kibana

             service kibana restart

How to extend the plugin

  - include Kibana's visualisations

 

 - provide user possibility to choose the index and fields

Further work

- extend the plugin to include visualisation for database alert and listener logs

 

- use machine learning to analyse them

 

- use Kibana's visualisations instead of D3 library

Thank you!