I currently lead the Engineering team at Facebook developing Cubrick, a new full-featured in-memory OLAP database for interactive analytics over highly dynamic datasets. Cubrick leverages a new indexing technique called Granular Partitioning, which I proposed in my PhD thesis, that allows the execution of indexed queries over any combination of dimensions on a multidimensional dataset, without incurring memory overhead (O(1) memory) and allowing constant time updates (O(1) time).
Prior to joining Facebook, I spent about 4 years working for the Brazilian Government, supporting the databases and infrastructure leveraged by the Brazilian Electoral Systems.
I received my BS, MS and PhD from the Federal University of Parana (UFPR) in Brazil.
Interests
Distributed systems, in-memory databases, large scale data processing and database indexing
Latest Publications
ICDE 2018 - April 16, 2018
Rethinking Concurrency Control for In-Memory OLAP DBMSs
Pedro Pedreira, Yinghai Lu, Sergey Pershin, Amit Dutta, Chris Crosswhite
VLDB - September 5, 2016
Cubrick: Indexing Millions of Records per Second for Interactive Analytics
Pedro Pedreira, Chris Croswhite, Luis Bona
VLDB 2015 - August 31, 2015
Cubrick: A Scalable Distributed MOLAP Database for Fast Analytics
Pedro Pedreira, Luis Erpen de Bona, Chris Croswhite
Latest News
