Added to README file links to documentation and an example to get the…

… raw data.

Change is related to issue #43.

README.rst

@@ -16,7 +16,7 @@ experimentation and tuning of data preparation as well as model parameters and
 hyper-parameters:
 
 * Cache raw data (dstat CSV files) and run metadata from OpenStack log server
-  and test result database. After this step, everything else may be done
+  and `test result database<https://docs.openstack.org/infra/system-config/logstash.html#subunit2sql>`. After this step, everything else may be done
   offline, as only cached data will be used
 * Build a normalized dataset, split into training, dev and test. This step
   allows to define the number of samples, the level of downsampling (in any)
@@ -33,7 +33,11 @@ To cache raw data locally, use `ciml-cache-data`:
 
   ciml-cache-data --build-name <build-name> --db-uri <db-uri>
 
-This connects to the OpenStack subunit2sql database, fetch all runs that
+Example:
+.. code::shell
+  ciml-cache-data --build-name tempest-full --db-uri mysql+pymysql://query:query@logstash.openstack.org/subunit2sql
+
+This connects to the OpenStack `subunit2sql<https://docs.openstack.org/subunit2sql/latest/reference/index.html>` database, fetch all runs that
 match the specified build_name and try to download the dstat data from
 logs.openstack.org. The dstat file and build results are stored gzipped
 under data/.raw and data/.metadata.