DBeaver

DBeaver Community / PRO — JDBC sessions

DBeaver is the cleanest demo of the Airbrx pitch. Change one field in the JDBC URL, run a query twice, and DBeaver's session log shows you the X-Airbrx-* headers right there — no App, no API call, nothing else needed to see the cache work.

Change the JDBC URL

Open the connection settings (Edit Connection → Connection settings) and update the host portion of the JDBC URL. Everything after the host stays the same.

Before

jdbc:snowflake://your-account
  .snowflakecomputing.com
  /?warehouse=REPORTING_WH
  &db=ANALYTICS

Direct JDBC to your warehouse.

After

jdbc:snowflake://your-slug
  .gateway.airbrx.ai
  /?warehouse=REPORTING_WH
  &db=ANALYTICS

Same JDBC, gateway address.

Databricks JDBC is the same idea — replace your cluster.cloud.databricks.com host with the gateway address and keep the rest of the URL (HTTP path, transport mode, SSL, etc.) unchanged.

Authentication

Use whatever auth your DBeaver connection used before — username and password in the dialog, key-pair via the private_key_file property, OAuth via the connector. The Gateway forwards the auth challenge to your warehouse without inspecting it.

Verify the cache — DBeaver shows the headers

Run any cacheable query twice. DBeaver's Session log captures the JDBC session metadata, including the X-Airbrx-* response headers Airbrx adds. The first run shows MISS; the second shows HIT:

# First run — warehouse executes, response cached X-Airbrx-Cache-Status: MISS X-Airbrx-Rule-Id: cache-aggregates-1h X-Airbrx-Execution-Time: 412ms # Second run, identical query — cached response X-Airbrx-Cache-Status: HIT X-Airbrx-Rule-Id: cache-aggregates-1h X-Airbrx-Execution-Time: 8ms

That side-by-side — 412ms drop to 8ms — is the Airbrx pitch in five seconds. Same query, same answer, same credentials, no warehouse compute the second time. Send the screenshot to your CFO.

Notes worth knowing

Driver compatibility. The Gateway is wire-compatible with Snowflake's and Databricks' official JDBC drivers. Use whatever driver version DBeaver bundles or downloads — no swap needed.
Connection pooling. DBeaver pools a small number of sessions per connection. The Gateway pools warehouse sessions independently, so this stacks fine — DBeaver opens a session to the Gateway, the Gateway opens (or reuses) a warehouse session.
Catalog browsing. When you expand the connection tree in DBeaver, it issues catalog / metadata queries. Those are excellent caching candidates — once cached, navigating the schema is instant.
SQL Editor execution times. DBeaver's status bar shows execution time per statement. After the cache warms, that number drops to milliseconds and stays there.

Where to go next

How the Gateway works — what the Gateway is doing for each query you ran.
Response headers reference — full list of X-Airbrx-* headers.
Other connection recipes.

Try it in DBeaver in five minutes.

Airbrx is in private preview. Create an account and run a query twice.

Create an account