Modern data clouds provide a managed data warehousing experience where users can quickly build their business intelligence applications and dashboards, while data cloud providers and distributed computing vendors take care of provisioning and scaling the data processing infrastructure on demand. Unfortunately, these data clouds are general purpose and are not optimized for the customer workloads at hand. In fact, data warehousing workloads have become too complex for customers to understand or tune manually. Prior on-prem database tuning tools relied on expert database administrators to identify bottlenecks, derive and apply the tuning decisions, monitor and roll them back in case of degradation, and constantly evolve the entire process over time. This is either too cumbersome or simply infeasible for modern data clouds.

Keebo introduces a data learning paradigm to automatically tune cloud data warehouses for better performance and lower cost. By continuously learning from the historical query workloads, Keebo simplifies the data engineering and helps achieve the business goals.

Publication

• Barzan Mozafari, Radu Alexandru Burcuta, Alan Cabrera, Andrei Constantin, Derek Francis, David Grömling, Alekh Jindal, Maciej Konkolowicz, Valentin Marian Spac, Yongjoo Park, Russell Razo Carranza, Nicholas Richardson, Abhishek Roy, Aayushi Srivastava, Isha Tarte, Brian Westphal, Chi Zhang
  Making Data Clouds Smarter at Keebo: Automated Warehouse Optimization using Data Learning
  SIGMOD, 2023, Seattle, USA.
  

Patents

• Alekh Jindal, Barzan Mozafari, David Wolfgang Groemling, Brian Westphal, Alan D Cabrera
  Managed tuning for data clouds.
  US Patent 11,693,857
• Alekh Jindal, Barzan Mozafari, Brian Westphal, Shi Qiao, Matthew Larsen, Advait Abhay Dixit
  Platform agnostic query acceleration.
  US Patent 11,567,936