#
# This project is no longer supported by Thumbtack Technology.
#

Yahoo! Cloud System Benchmark (YCSB)
====================================

This version of the YCSB tool is modified by Thumbtack Technology to add 
support for Aerospike and Couchbase databases, to improve MongoDB driver 
and to add some automation to run YCSB on multiple clients.

Links
-----
http://wiki.github.com/brianfrankcooper/YCSB/
https://github.com/couchbaselabs/YCSB
http://research.yahoo.com/Web_Information_Management/YCSB
ycsb-users@yahoogroups.com
http://docs.fabfile.org/en/1.3.2/

Getting Started
---------------

1. Download the latest release of YCSB:

    git clone https://github.com/thumbtack-technology/ycsb.git
    cd ycsb

2. Set up a database and client hosts to benchmark. 
   There is a README file under each binding directory.
   
   You must have SSH (and in most cases root) access to all your hosts.

3. Configure YCSB build script to build database binders
   Edit pom.xml, uncomment modules related to databases which you chose in <modules> section

4. Configure hosts, databases and workloads settings:
   Edit files: conf/hosts.py, conf/databases.py, conf/workloads.py

5. Build and deploy YCSB to client hosts

    fab ycsb_deploy

6. Load data to databases

    fab ycsb_load:db=<dbname>

   <dbname> is aerospike, couchbase, couchbase2, cassandra, 
   mongodb or any other you configured
   Edit conf/workloads.py to confiture workloads root directory

7. Run YCSB workload

    fab ycsb_run:db=<dbname>,workload=A

8. Check the YCSB status
 
    fab ycsb_status:db=<dbname>

9. Download YCSB results and logs

    fab ycsb_get:db=<dbname>,do=True
    
   You'll get some .out and .err files in the current directory downloaded
   from all your clients.

10. Aggregate the YCSB results

    ./bin/merge.py
    
   This script gets the most important parameters from YCSB .out files,
   such as throughput and latency, aggregates the results from multiple clients
   and prints the result as tab-separated values which can be easy pasted
   into any spreadsheet.

Notes
-----

This tool was tested using following software versions
* Ubuntu Server (12.04)
* Git (1.7.10.4)
* openjdk-7-jdk (7u9-2.3.3)
* Maven (2.2.1)
* Fabric (1.3.2)
* Python (2.7.3)