The Trip to The Enterprise Gourmet Data Product Marketplace through a Self-service Data Platform

Abstract: Data Analytics provides core business reporting needs in many software companies, acts as a source of truth for key information, and enables building advanced solutions, e.g., predictive models, machine learning, real-time recommendations, to grow the business. A self-service, multi-tenant, API-first, and scalable data platform is the foundational requirement in creating an enterprise data marketplace, which enables the creation, publishing, and exchange of data products. Such a marketplace enables the exploration and discovery of data products, further providing high-level data governance and oversight on marketplace contents. In this paper, we describe our way to the gourmet data product marketplace. We cover the design principles, the implementation details, technology choices, and the journey to build an enterprise data platform that meets the above characteristics. The platform consists of ingestion, streaming, storage, transformation, schema generation, fail-safe, data sharing, access management, PII data automatic identification, self-service storage optimization recommendations, and CI/CD integration. We then show how the platform enables and operates the data marketplace, facilitating the exchange of stable data products across users and tenants. We motivate and show how we run scalable decentralized data governance. All of this is built and run for Cimpress Technology (CT), which operates the Mass Customization Platform for Cimpress and its businesses. The CT data platform serves 1000s of users from different platform participants, with data sourced from heterogeneous sources. Data is ingested at a rate of well over 1000 individual messages per second and serves more than 100k analytical queries daily.