Getting Started

YCSB is a framework for benchmarking systems. By itself, it is not particularly useful; only when you add code to interface with a data serving system is it useful. Each YCSB release contains code to interface with various DB bindings, here are some examples:

• Accumulo
• Cassandra
• Couchbase
• DynamoDB
• ElasticSearch
• HBase
• HyperTable
• Infinispan
• JDBC
• MongoDB
• OrientDB
• Redis
• Tarantool

Refer to the release notes of the latest version for a complete list of DBs supported.

It is straightforward to interface with other database systems - see Adding a Database.

1. Obtain YCSB

If you'll be running on Windows, please start by referencing our prerequisites for Windows.

Download the latest version:

curl -O --location https://github.com/brianfrankcooper/YCSB/releases/download/0.17.0/ycsb-0.17.0.tar.gz
tar xfvz ycsb-0.17.0.tar.gz
cd ycsb-0.17.0

Or clone the git repository and build:

git clone git://github.com/brianfrankcooper/YCSB.git
cd YCSB
mvn clean package

Systems may have additional requirements for running clients. For example, HBase requires the client be able to contact Zookeeper. See HBase Fully Distributed Quickstart Guide for HBase-specific instructions. Some details are listed in Using the Database Libraries.

You will be using the ycsb command to interact with YCSB. Run:

./bin/ycsb

to see the usage.

2. Now, run a workload

See Running a Workload.