efficient/cicada-exp-sigmod2017-DBx1000

Name: cicada-exp-sigmod2017-DBx1000

Owner: Efficient Computing at Carnegie Mellon

Description: A fork of DBx1000 for Cicada SIGMOD 2017 evaluation

Created: 2016-05-07 00:07:36.0

Updated: 2017-10-29 11:53:12.0

Pushed: 2017-06-09 21:34:36.0

Homepage:

Size: 651

Language: C++

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

DBx1000

DBx1000 is an single node OLTP database management system (DBMS). The goal of DBx1000 is to make DBMS scalable on future 1000-core processors. We implemented all the seven classic concurrency control schemes in DBx1000. They exhibit different scalability properties under different workloads.

The concurrency control scalability study is described in the following paper.

Staring into the Abyss: An Evaluation of Concurrency Control with One Thousand Cores
Xiangyao Yu, George Bezerra, Andrew Pavlo, Srinivas Devadas, Michael Stonebraker
http://www.vldb.org/pvldb/vol8/p209-yu.pdf

Build & Test

To build the database.

make -j

To test the database

python test.py

Configuration

DBMS configurations can be changed in the config.h file. Please refer to README for the meaning of each configuration. Here we only list several most important ones.

THREAD_CNT        : Number of worker threads running in the database.
WORKLOAD          : Supported workloads include YCSB and TPCC
CC_ALG            : Concurrency control algorithm. Seven algorithms are supported 
                    (DL_DETECT, NO_WAIT, HEKATON, SILO, TICTOC) 
MAX_TXN_PER_PART  : Number of transactions to run per thread per partition.

Configurations can also be specified as command argument at runtime. Run the following command for a full list of program argument.

./rundb -h

Run

The DBMS can be run with

./rundb