With a PhD in Computer Science and Engineering from the Hong Kong University of Science and Technology, Papadopoulos worked as a research scientist at Massachusetts Institute of Technology and Intel Labs prior to launching TileDB. As he explains in this interview, the idea for TileDB originated in that research work in emerging big data systems and the hardware requirements to support those workloads.
Universal databases are not new, but they are re-emerging as an alternative to the single-purpose databases that have become popular in the tech industry.
Key topics from the interview include:
- TileDb stores data in multi-dimensional arrays, or matrixes. The data types and workloads it supports.
- How TileDB differs from object-relational universal databases of a generation earlier.
- How TileDB compares to purpose-built databases – time-series, graph, document, vector, etc.
- Use cases and early adopters.
- TileDB’s availability as a cloud service and for use on-premises.
Quotes from the podcast:
- “These ideas were shaped based on interactions we had with practitioners and data scientists across domains. That was key. We did not delve into the traditional, relational query optimization and SQL operations that other people were doing with different architectures in the cloud."
- "I was very drawn to scientific use cases like geospatial and bio-informatics. And it came as a great surprise to me that none of those verticals and applications were using databases."
- "Is there a way to build a single storage engine to consolidate this data? A single authentication layer, a single access control layer, and so on. This is how it started."