Global Top 10 Financial Institution Reduces Compute Waste and Simplifies Analytics

Overview

A leading global financial institution, consistently ranked among the world’s top 10, operates across multiple divisions, each producing curated, high-quality datasets to support enterprise-wide reporting, governance, and compliance. The enterprise governance and shared services function consumes these datasets to support regulatory alignment and cross-functional analytics across Finance, Risk, Compliance, Legal, IT, HR, and Operations.

As data volumes grew, each division maintained its own Databricks cluster to ensure isolation and compliance. While effective for governance, this model led to underused consumption clusters, rising operational costs, and increasing complexity. To address this, the enterprise deployed Denodo’s Embedded MPP engine as a shared analytical compute layer, consolidating reporting workloads while keeping Databricks focused on data preparation, transformation, and ML.

Why Denodo

Denodo’s Embedded MPP engine provides a shared analytical compute layer while preserving strong governance controls.

Denodo enables the institution to:

• Consolidate dashboard and reporting workloads on one shared MPP engine rather than using multiple Databricks clusters
• Maintain tenant isolation and compliance via Denodo’s policy-based access and tagging framework
• Run governed, cross-source analytical queries across cloud and on-premise systems from one logical layer
• Keep Databricks focused on advanced data engineering and ML, while Denodo MPP powers cost-efficient analytical consumption

Challenges

Rising data volumes and redundant compute highlighted key challenges in the institution’s analytical architecture.

• Each tenant maintained its own Databricks cluster to query Delta data, ensuring isolation and compliance but creating a highly redundant setup
• Many of these clusters were used primarily for lightweight, ad hoc dashboard and reporting workloads, which did not require the full power of a Spark-based environment
• Maintaining and supporting multiple clusters, along with the pipelines that connected them through Denodo, became increasingly complex and costly

Solution

The institution evaluated Denodo’s Embedded MPP engine as a shared analytical compute layer to consolidate reporting workloads without disrupting Databricks’ engineering functions. Production workloads were tested across both platforms using Delta tables in ADLS, confirming that Denodo MPP could support dashboard queries at comparable performance while removing cold-start delays and reducing the need for multiple consumption clusters.

With this validation, the organisation adopted Denodo MPP as the central layer for analytical consumption, while continuing to use Databricks for data preparation, transformation, and machine learning.

Key Benefits

The introduction of Denodo’s Embedded MPP engine simplified analytical consumption while maintaining the governance and performance standards required by a global financial institution.

• One shared Denodo MPP cluster replaced multiple Databricks consumption clusters, reducing compute waste and operational overhead
• Up to USD 150,000 in annual Databricks compute savings were identified from Denodo-related workloads
• Logical tenant isolation and policy-based access ensured security and compliance without infrastructure duplication
• The governance function now manages one scalable compute layer, while Databricks remains dedicated to high-performance engineering and ML

Click here to download the complete case study.