News & Events
Carnegie Mellon Database Group Wins 2018 IEEE ICDM Test-of-Time Paper Award
Singapore - The Carnegie Mellon Database Group is pleased to announce that their 2009 paper PEGASUS: A Peta-Scale Graph Mining System - Implementation and Observations has won the 2018 IEEE ICDM Test-of-Time Award. The authors were CMU Ph.D. students U Kang and Charalampos Tsourakakis, in collaboration with Prof. Christos Faloutsos. This paper on the PEGASUS project showed how to apply graph-mining algorithms on a Map-Reduce platform. The main insight was that a wide range of graph mining algorithms eventually require Read More
[DB Seminar] Fall 2018: Ethan Zhang (VoltDB)
Following from the idea that "one size no longer fits for all", a family of "NewSQL" specialized databases arose. To handle OLTP, researchers at MIT and Brown (and a few other places) built H-Store, a distributed, shared-nothing, in-memory database that got rid of locking, latching, buffering, and logging, beating the performance of traditional OLTP RDBMSs by nearly two orders of magnitude. This prototype was commercialized in 2009 as VoltDB. Since then, the two have lived full and productive lives in Read More
[DB Seminar] Fall 2018: Yihan Sun
Modern query-heavy applications of database systems especially require minimal delays to OLAP queries, as well as allowing the lasted OLTP updates to be visible in time. A popular mechanism for fast response to OLAP queries is to use snapshot isolation (SI) for multi-version concurrency control (MVCC), as it allows readers to make progress regardless of concurrent writers. Many other optimizations for OLAP queries include denormalization, materialization view and table partitioning. However, most existing solutions to these optimizations do not support Read More
[DB Seminar] Fall 2018: Lin Ma
In this talk, I will present the progress on the self-driving database project. This is a 10-min practice talk for the PDL retreat. Read More
[Hardware Accelerated Databases] Richard Heyns (Brytlyt)
In this talk, we will cover how the implementation of GPU database management systems are different than CPU database systems and provide evidence that shows how much of the performance gains with these systems are achieved via just GPUs. We will also discuss how we are solving the problems of tomorrow – making AI smarter, faster and more intuitive with Brytlyt's BrytMind by combining SQL with its GPU Manager and AI. We will also explore the different ways GPU Databases Read More
[DB Seminar] Fall 2018: System Design Planning II
We will discuss the ongoing work on porting the LLVM engine to the new system. Read More
[Hardware Accelerated Databases] Karsten Rönner (Swarm64)
Online Analytical Processing (OLAP) of very large data sets and/or high-velocity data is a workload that strains all parts of a compute system: storage bandwidth, IO-subsystem throughput, main-memory bandwidth, instruction-level concurrency and thread-parallelism. Swarm64 seeks to improve the effective throughput and the compute efficiency of OLAP workloads by adding FPGAs as additional compute element to standard compute servers. The hard- and software stack performing OLAP workloads, the S64 DA, has been designed to integrate into popular SQL open-source databases through Read More
[DB Seminar] Fall 2018: System Design Planning
More design discussion of the new DBMS. Read More
[Hardware Accelerated Databases] Felipe Aramburu (BlazingDB)
BlazingDB has spent the past six months working on an open-source project (libgdf) alongside Anaconda and Nvidia. Libgdf is a library of computational primitives on top of a memory layout which is similar to Apache Arrow but optimized for GPUs. We have created a distributed, GPU-accelerated ETL pipeline that takes a user from reading data in Parquet, to performing SQL operations over that dataset, and finally feeding that data into xgboost, a machine learning library that allows us to leverage Read More
[DB Seminar] Fall 2018: System Design Planning with Prashanth Menon
This meeting will be entirely about discussing the design and implementation of CMU's new DBMS. The focus will be on how we combine the new storage manager with the beautiful Prashanth Menon's LLVM execution engine. Read More