Heading towards Data Mesh for Orange, which is migrating to a lakehouse with Starburst


It is not only in France that Orange can declare itself “data rich”. In Ivory Coast, where its subsidiary is positioned as the leading operator with more than 17 million active customers, Orange also has significant data assets.

In the country, the multinational has carried out a project to modernize its data infrastructure. In this context, Orange Côte d’Ivoire deployed the Starburst Enterprise data virtualization solution.

From a Hadoop Data Lake to a Lakehouse

Presenting itself as a data lake analytical platform, the American publisher’s technology should enable it to streamline its operations and improve accessibility to data stored in multiple sources.

According to Starburst’s press release, the use of its platform was at the origin of a migration of data to a lakehouse – a cross between data lake and data warehouse. The migration was based on use cases.

“By decoupling compute performance from storage, and taking advantage of new technologies implemented, the company was able to create a modern, multi-petabyte lakehouse.”

In addition to gains in terms of accessibility, the operator aimed to prevent a risk of overdependence on one supplier. Orange also wanted to be able to manage the most common open table and file formats.

More autonomous professions in access to data

Also at stake is “more autonomy for business teams” in the use of data, and in particular for the fight against fraud. This empowerment is enabled by access to SQL, which “has accelerated time-to-insight for Data Analysts.”

The Boston-based publisher also highlights the benefits of this implementation for Data Engineers. These functions are “freed from repetitive extraction tasks and can be mobilized on projects with higher added value”, claims Starburst, which cites in particular the development of their skills in artificial intelligence.

In the old architecture, Data Engineers were forced to process numerous data extraction requests from business lines, with the consequence of creating a bottleneck internally and extending processing times.

Additionally, Orange “also did not have advanced skills to modernize its existing Hadoop Data Lake, which posed challenges in terms of performance and governance.”

Data Mesh in hybrid environment

“Starburst has truly become our single point of access to data, instead of having to send requests to different sources separately,” explains Armand Kablan, Data & AI Manager at Orange Côte d’Ivoire.

To illustrate the added value of its technology, the company particularly highlights “the considerable time savings in processing data analysis requests”. The ROI: a 300% increase in productivity.

“For compliance, requests that previously took several hours on Hadoop now only take a few minutes,” the publisher further indicates. Other steps now await the Orange subsidiary.

Presented as “the cornerstone of the data management strategy” and “its main ally for setting up Data Products and a Data Mesh architecture”, Starburst Stargate will be deployed to bring together on-prem and cloud. Orange also plans to integrate data on Big Query and locally.

To move to a decentralized Data Mesh type logic, Orange Côte d’Ivoire operates a hybrid infrastructure based on Google Cloud Platform (GCP) and an on-premise Cloud Native platform composed of Dell servers.



Source link -97