Create a PostgreSQL-compatible JDBC connection so Looker can query ParadeDB datasets.
Connecting ParadeDB to Looker lets analysts build real-time dashboards on ParadeDB’s vector and relational data without ETL. ParadeDB speaks the PostgreSQL wire protocol, so Looker treats it as a standard Postgres database.
In Admin → Connections choose Dialect PostgreSQL. Enter host, port, database, username, and password. ParadeDB usually listens on port 5432.
Set SSL = require unless ParadeDB runs on the same private network. ParadeDB supports the standard sslmode parameters.
Looker builds the string from the form fields, but for reference it resolves to jdbc:postgresql://HOST:PORT/DATABASE?sslmode=require
. ParadeDB requires no extra parameters.
Create a dedicated ParadeDB role with SELECT privileges on reporting schemas. Deny INSERT/UPDATE/DELETE to prevent accidental writes.
CREATE ROLE looker_readonly LOGIN PASSWORD '•••';GRANT CONNECT ON DATABASE ecommerce TO looker_readonly;GRANT USAGE ON SCHEMA public TO looker_readonly;GRANT SELECT ON ALL TABLES IN SCHEMA public TO looker_readonly;ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT SELECT ON TABLES TO looker_readonly;
Click Test These Settings in Looker. A green check confirms connectivity. Run a sample explore to verify query results.
Create a LookML model sourcing the Customers
, Orders
, and OrderItems
tables. Join on customer_id
to visualize revenue per customer.
Enable connection pooling in ParadeDB’s postgresql.conf
for high-concurrency BI workloads. Monitor slow queries with pg_stat_statements
and add indexes on join columns like Orders.customer_id
.
"FATAL: no pg_hba.conf entry" means ParadeDB’s pg_hba.conf
blocks Looker’s IP; add a hostssl line. "password authentication failed" indicates a bad credential or wrong database.
No. ParadeDB is protocol-compatible with PostgreSQL 14+. Looker’s built-in driver works out of the box.
Yes, but only through SQL. Build Explore fields that call functions like vector_distance
. Looker will pass the SQL to ParadeDB unchanged.
Increase ParadeDB’s work_mem
for complex joins, add partitioning on large fact tables, and deploy replicas to distribute read traffic.