Speaker:

Nuno Faria

Date:

Tue Sep 8, 2026 @ 12:00pm EDT

Date:

Tue Sep 8, 2026

Time:

12:00pm EDT

Location:

GHC 8115

Title:

Bridging the Gap Between HTAP and Lakehouses for Analytics on Fresh Data

---

Talk Info:

There has been a growing interest in systems that combine analytical performance with some form of transactions to support ever more frequent updates. On the one hand, hybrid transactional analytical processing (HTAP) systems support fine grained updates at the expense of using custom data formats and processing engines or of fully replicating data between transactional and analytical subsystems, thus limiting scalability. On the other hand, lakehouse systems are limited to coarse grained updates but store data in open file formats such as Parquet, often in cloud-based object-stores, and support processing with datalake engines such as DataFusion. In this talk, we discuss how to build a system whose transactional performance competes with HTAP (e.g., on TPC-C) while retaining the efficiency, openness, and modularity of lakehouse systems. In detail, we focus on how Parquet, DataFusion, and PostgreSQL can be reused for this purpose.

Bio:

Nuno Faria is a researcher at INESCTEC and U. Minho. He is a member of the board at the Graph Data Council and an Apache DataFusion Committer.