jason w. jones

software from the trenches

Thriller

Thriller is a JDBC driver that runs MDX queries against an Essbase server and hands the results back as an ordinary, typed result set. That lets anything that takes a JDBC driver work with a cube: IDEs such as IntelliJ and DataGrip, database tools such as DBeaver and RazorSQL, sqlline in a terminal, reporting and integration tools, or your own Java code.

It’s open source under the Apache 2.0 license, and the code is on GitHub. Thriller talks to Essbase over the REST API, through essbase-rest-client, so it needs no Oracle software. It is tested against Essbase 26.1, and its first release against 21.1 too. Thriller 1.1 has the story of how it got here.

Getting It

For a SQL tool, download thriller-mdx-jdbc-1.1.0-all.jar from the releases page or Maven Central. It’s a single jar with everything Thriller needs. Add it to the tool as a driver, with the class com.jasonwjones.hyperion.thriller.ThrillerDriver.

For code, it’s on Maven Central:

<dependency>
    <groupId>com.jasonwjones.hyperion.thriller</groupId>
    <artifactId>thriller-mdx-jdbc</artifactId>
    <version>1.1.0</version>
</dependency>

Connecting

The JDBC URL is the Essbase server’s address, as you’d open it in a browser, and the user and password are your Essbase login:

String url = "jdbc:essbase:thriller:http://localhost:9000/essbase";
Connection conn = DriverManager.getConnection(url, "admin", "password");

Each query reads the cube its FROM clause names. Settings such as ;members=alias can follow the address in the URL; the README lists them all.

Shaping the Results

Without any help, Thriller names the columns from the result: one for each dimension on the rows, holding member names, then one for each data column, named for the members above it. To choose the names and types yourself, put a block of hints just after SELECT:

SELECT
/*
 -- column.1.name = MEASURE
 -- column.1.type = VARCHAR
 -- column.2.name = JAN
 -- column.2.type = DECIMAL
 -- column.2.scale = 2
 */
{[Jan]} ON COLUMNS, {[Measures].Levels(0).members} ON ROWS FROM [Sample].[Basic]

Hints don’t have to cover every column: -- column.2.scale = 2 alone would just round the second. Other hints name the members by alias (-- members = alias), turn the grid into a fact table with a row per value (-- shape = long), and add the point of view as columns (-- pov = columns).

The block goes after SELECT because some tools, IntelliJ among them, run only the statement under the cursor, and a comment above it isn’t part of the statement.

In Your Tools

Tools that show a database as a tree see each application as a schema, each cube as a table, and each dimension as a column, along with the server’s MDX functions. In IntelliJ’s console, use its ${name} placeholders to fill in values, such as {[${product}]}.

The cookbook has twelve queries against Sample.Basic, each with the result Thriller returns for it.

History

Thriller started in 2016 as an experiment on top of the Essbase Java API: Running MDX queries through a JDBC driver (for fun?), then Playing with the Thriller MDX/JDBC Driver in Drillbridge and an update that November. In 2026 it was rewritten over the REST API and released as open source.