Connect Amazon Redshift to Dataki

02·Set it up

3 steps, about 10 minutes

1. 01

   Create a read-only user

   Connect as the admin user, for instance in Query Editor v2, and create a user for Dataki. Pick a password of 8 to 64 characters with an upper-case letter, a lower-case letter and a digit, and without ', ", \, / or @. Then give it USAGE on each schema Dataki should read and SELECT on the tables in it.

   The last line is Redshift's scoped permission: it also covers tables created in the schema later, by anyone. Where it is not available to you, the line before it grants the tables that exist today, and you run it again after adding tables.

   CREATE USER dataki PASSWORD 'Choose-a-long-password-1';
   GRANT USAGE ON SCHEMA analytics TO dataki;
   GRANT SELECT ON ALL TABLES IN SCHEMA analytics TO dataki;
   GRANT SELECT FOR TABLES IN SCHEMA analytics TO dataki;

   • Repeat the three GRANT lines for every schema you want to ask about.
   • Set the password as plain text, as above. A user whose password was set as a sha256|… hash cannot sign in from Dataki.
2. 02

   Let Dataki in

   Redshift creates clusters and workgroups private, and new ones accept only TLS. For Dataki to reach yours it has to be publicly accessible, with an inbound rule for 34.89.253.13.

   • Network settings

     Provisioned cluster

     The cluster's Properties tab › Network and security settings.

     Serverless workgroup

     Workgroup configuration › your workgroup › Data access › Network and security › Edit.

   • Security group rule

     Provisioned cluster

     Open the VPC security group › Inbound rules › Edit inbound rules: TCP, port 5439, source 34.89.253.13/32.

     Serverless workgroup

     The same: TCP, port 5439, source 34.89.253.13/32.

   • Public access

     Provisioned cluster

     Actions › turn on Publicly accessible.

     Serverless workgroup

     Turn on Publicly accessible in the same panel.

   • Endpoint

     Provisioned cluster

     General information › Endpoint.

     Serverless workgroup

     The workgroup's General information › Endpoint.

   	Provisioned cluster	Serverless workgroup
   Network settings	The cluster's Properties tab › Network and security settings.	Workgroup configuration › your workgroup › Data access › Network and security › Edit.
   Security group rule	Open the VPC security group › Inbound rules › Edit inbound rules: TCP, port 5439, source 34.89.253.13/32.	The same: TCP, port 5439, source 34.89.253.13/32.
   Public access	Actions › turn on Publicly accessible.	Turn on Publicly accessible in the same panel.
   Endpoint	General information › Endpoint.	The workgroup's General information › Endpoint.

   • The VPC needs a route to an internet gateway, which the default VPC has. When public access is on, Redshift gives the cluster or workgroup an Elastic IP of its own.
3. 03

   Add the connection in Dataki

   Open app.dataki.ai/connect and, under Database Connections, choose Connect Redshift. Paste the endpoint as the console shows it, …amazonaws.com:5439/dev: Dataki takes the port and the database from it. Enter dataki and its password, then Test connection and Save Connection.

   Leave Schemas empty for every schema the user can read, or list the ones the model should see, separated by commas.