Speaker:

Ippokratis Pandis

Date:

Mon Oct 26, 2026 @ 04:30pm EDT

Date:

Mon Oct 26, 2026

Time:

04:30pm EDT

Location:

https://cmu.zoom.us/j/92568565952?pwd=N8VKf5xfF5aisG9VoEwdNsCp0dCwel.1Zoom

Title:

Lakebases, Lakehouses, and LTAP: Unifying Transactions and Analytics on Open Storage

System:

Databricks

---

Talk Info:

Traditionally, data architectures maintain a strict division between high-throughput operational databases and large-scale analytical lakehouses. The general wisdom for organizations seeking low-latency real-time analytics is to copy data into dedicated serving layers, which creates data redundancy, stale insights, and fragmented governance.

In this talk, we explore how unifying operational and analytical workloads on open storage eliminates the need to copy data. In the first part, we present LakehouseRT, an analytical query engine designed to deliver millisecond-level query latency and high concurrency directly over open table formats in object stores governed by a unified catalog. In the second part, we show how Lakebase and its disaggregated compute and storage OLTP architecture not only provides best-in-class performance, elasticity, and availability for Postgres workloads, but also seamlessly enables low-latency analytics directly on live operational data without compromises.

This talk is part of the Whatever Andy Found Interesting While Out At Sea.

Bio:

Ippokratis Pandis is a Distinguished Engineer at Databricks working on data management systems. Previously, he spent a decade at AWS, where he served as Chief Architect of AWS Analytics and led the development of Amazon Redshift. Earlier, he contributed to Cloudera Impala and IBM DB2 BLU. He earned his Ph.D. from Carnegie Mellon University, where in the dungeons of Wean and Hamerschlag Halls, he worked on multicore transaction processing. He has authored numerous research papers and served the database community as General Chair of SIGMOD 2023 and President of HPTS.