Prologis manages a wide breadth of real estate assets across the globe, which translates into a staggering amount of data that is stored across geographically dispersed sources. Every day, Prologis needs to integrate this vast amount of data, just to operate its business effectively.
For many years, Prologis managed its data via a global, on-premises data warehouse comprised of 27 servers supporting a series of databases, integration servers, and reporting servers. Dedicated systems captured changed data from source systems and sent it to the data warehouse using extract, transform, and load (ETL) scripts. Prologis became aware of the disadvantages of its infrastructure, as ETL scripts need to be rewritten and re-tested whenever there is a change in the source environment.
Growing Pains
“Six years ago, this was modern technology, but we began to outgrow it,” says Luke Slotwinski, VP, IT Data and Analytics at Prologis. Prologis wanted to modernize its data infrastructure to include cloud capabilities, as well as to introduce efficiencies that would accelerate advanced analytics. However, Prologis was reluctant to engage in modernization efforts that would cause undue downtime that would result from interruptions to the existing system. Specifically, the Prologis data science team was working on a cost optimization use case. but the team was spending a majority of the project time in data exploration, ingestion, and manipulation tasks, leaving very little time to focus on creating and training their data models. “The biggest problem of advanced analytics is that 80% of the time is spent in cleansing, analyzing, and massaging data, and only 20% on creating and running the math model,” says Slotwinski.
Smooth Modernization
In order to modernize and futureproof its data infrastructure, Prologis leveraged the Denodo Platform to integrate data in real time from myriad disparate data sources without having to replicate data to a new consolidated repository. The Denodo Platform acts as an intelligent data-access layer between data sources and data consumers, abstracting data consumers from the complexities of access. This enabled Prologis to establish a logical data warehouse architecture that provides real-time access to data across on-premises, cloud, and other sources. The virtual layer acts as a data delivery layer for various advanced analytics programs aggregating data from multiple sources and feeding the resultant set to the analytical models. For Prologis data scientists, this approach saved a lot of time normally spend on upfront data exploration and manipulation.
The Denodo Platform helped Prologis transition to a cloud based platform without affecting the business continuity. “While the migration took place,” explains Slotwinski, “users and reporting tools were still talking to the Denodo Platform, while behind the scenes, our team was biting chunks away from our old architecture and putting those chunks into the new architecture, seamlessly moving down that path.” Once the migration was complete, and Prologis implemented a logical architecture, the company began to appreciate the Denodo Platform’s ability to accelerate analytics to support a broad array of use cases. “Earlier, for a new table that we’ve never analyzed before, we needed to dedicate a sprint to get the ETL set up to get the data into a database so we could start analyzing it,” says Slotwinski. “Now, we can virtualize that table and immediately start doing analytics on it.” Slotwinski estimates that this accelerated analytics activities by as much as 30%.
To learn more, see the case study. Or see all case studies across a variety of industries and use cases.
