Covestro uses AWS to transform and speed up operations

Premium polymer supplier Covestro has customers across the world and many companies in industries including automotive, construction, healthcare, energy and electronics rely on its high-tech materials. To differentiate its business, Covestro decided to transforming its IT organisation several years ago. However, to do that it had to migrate hundreds of on-premises business applications from data centres to the cloud.

Michael Schuster, VP and Head of Application Services, Covestro said: “We wanted to move to the cloud to create a more modern, agile IT department that could become more of a value provider for business units like research and development. We have a lot of data, and we wanted to transform it into value through machine learning and then feed insights back to the business.”

Covestro chose Amazon Web Services (AWS) for its migration, which would also enable the company to make use of AWS Customer Enablement and AWS Professional Services.

“We needed a technology provider that understood our business culture, and AWS was the best fit,” Schuster says. “In addition, AWS understands that the cloud isn’t only about moving out of a data centre. It’s also about changing an entire operating model within the IT organisation,” Schuster said.

Covestro worked closely with AWS Professional Services consultants to move over 500 business applications and 1,000 servers from its data centres to an AWS Landing Zone. AWS Professional Services then helped Covestro build its new data self-service analytics solution, Covestro Analytics Platform, that uses machine learning (ML) technology. It also now uses Amazon SageMaker to build and train ML models, and AWS Glue to ingest and visualise data from SAP and other applications.

Now that Covestro runs its applications on AWS, it can provision capacity and storage in several minutes, where it used to take up to eight weeks in its previous data centre environment. It also creates and tests new business initiatives faster before moving them into production.

“What would traditionally take several weeks because of the need to procure and set up hardware now only takes a few hours,” says Schuster. “We can benchmark projects faster now and make data-driven decisions on which ones to move forward with.”