Optimizing hybrid architecture for greater data access and accelerated analytics

About:

Crédit Mutuel Arkéa (CMA), one of France’s top five banks with a customer base exceeding 30 million, operates as a mission-driven company, prioritizing social and environmental impacts. With a diverse portfolio encompassing banking, insurance, asset management, real estate, financial services, and payment services, the group’s 40 subsidiaries span five European countries. The group counts more than 11,000 employees worldwide, and more than 5 million customers and members.

Challenge: 

Facing exponential data growth and increasing analysis demands, Crédit Mutuel Arkéa encountered challenges in its data architecture. These challenges included:

• Slow query performance with Hive.
• Inability to connect BI tools directly to the on-premises data lake (Hadoop).
• Heavy dependence on ETL and batch processes, hindering data availability.
• Security vulnerabilities in data access and the inability to federate data across on-prem and cloud.

Recognizing that a full-scale data migration to the cloud might not be feasible due to security concerns and the lengthy, costly nature of such projects, Maxime Havez, Chief Data Officer at Crédit Mutuel Arkéa, envisioned a data architecture that favored hybridization. This vision sought a middle ground between purely on-premises and cloud-based solutions.

Solution: 

In response to these challenges, and guided by the vision of hybridization, Crédit Mutuel Arkéa chose Starburst Enterprise as its federated query engine. The decision was driven by:

• Starburst’s alignment with CMA’s preference for SQL and open source technologies.
• Governance model compatibility and quick testing.
• Starburst functionalities that resonated with their needs (i.e. federation, batch processing, and NoSQL database support).
• Starburst’s ability to deliver the required performance for querying massive data volumes spanning multiple petabytes.

CMA had considered alternative vendors, one being Denodo; however the company’s values did not align with the bank’s strategic priorities or preference for open source technology.

Results: 

Deploying Starburst as the data consumption and federation layer for CMA’s hybrid data architecture brought about the following results:

• Queries became 10 to 20 times faster, significantly reducing time-to-insight and enhancing productivity.
• Internal user satisfaction surged, with the user base increasing exponentially from 10 to 550 within a few months.
• Users, including non-Hadoop users, gained access to previously unavailable data, fostering self-sufficiency in data analysis. “These users include not only IT professionals but also many business users (65%) who appreciate the user-friendly nature of Starburst,” shares Havez.
• The data office team is more inclined to embark on new projects and explore other use cases, actively contributing to the company’s innovation initiatives.

“Starburst was quickly integrated into the information system (IS) of Arkea’s BigData platform, and perfectly fulfilled its role of accelerating SQL queries on the Hadoop infrastructure. Starburst makes it possible to hybridize the IS and centralize access control, enabling us to envisage an adapted and facilitated Data Lake evolution,” summarizes Anne Despoulains, Infrastructure Department Manager, Crédit Mutuel Arkéa.

Crédit Mutuel Arkéa’s journey with Starburst not only resolves immediate challenges but also paves the way for a future-proof, hybrid approach to data management.

Starburst’s technology enables the bank to seamlessly integrate data in Google Cloud Platform (GCP) with an on-premises cluster, while also allowing them to scale up and down according to its needs. The implementation unlocks new possibilities for analytics and innovation while ensuring a secure data integration across diverse environments.

More resources: Stéphane Orieux’s Data Rebel Profile and Maxime Havez’s Data Rebel Profile