Skip to main content

Perform a stress test using nosqlbench

Nosqlbench is an open source project that can be used to stress-test and benchmark several SQL and NOSQL databases including Cassandra® and PostgreSQL®.

Prerequisites

To download the latest nosqlbench release, search for "latest" in nosqlbench GitHub repository. Nosqlbench can be downloaded as a Linux binary executable called nb. The nb executable is built with all java libraries and includes a number of sample scenarios ready to be run.

tip

You can read more about the nosqlbench core concepts and parameters in the dedicated documentation

Variables

These are the placeholders you will need to replace in the code sample:

VariableDescription
PASSWORDPassword of the avnadmin user
HOSTHost name for the connection
PORTPort number to use for the Cassandra service
SSL_CERTFILEPath of the CA Certificate for the Cassandra service
tip

All the above variables and the CA Certificate file can be found in Aiven Console > your service's Overview page > Connection information section.

Run nosqlbench against your Aiven for Apache Cassandra® service

The following sections shows how to run several nosqlbench workloads against an Aiven for Cassandra service.

Create a schema and load data

Nosqlbench can be used to create a sample schema and load data.

The schema can be created with the following command, after substituting the placeholders for HOST, PORT, PASSWORD and SSL_CERTFILE:

./nb run                            \
host=HOST \
port=PORT \
username=avnadmin \
password=PASSWORD \
localdc=aiven \
driver=cql \
workload=cql-keyvalue \
ssl=openssl \
certFilePath=SSL_CERTFILE \
tags=block:schema

The following parameters are used:

  • driver: specifies the type of client you are going to use, in the example cql
  • workload: option calls a specific workload file that is compiled inside the nb executable and instructs nb to generate key/value pairs for a table called baselines.keyvalue. You can read more on how to define custom workloads in the dedicated documentation
  • block: refers to a specific point in the workload definition file and specifies the particular activity to run. In the example, the block is schema which means that the nosqlbench will create the schema of the Cassandra keyspace.

To create client connections and produce data in the keyspace and tables created, you need to run the following command line, after substituting the placeholders for HOST, PORT, PASSWORD and SSL_CERTFILE:

./nb start \
host=HOST \
port=PORT \
username=avnadmin \
password=PASSWORD \
localdc=aiven \
driver=cql \
workload=cql-keyvalue \
ssl=openssl \
certFilePath=SSL_CERTFILE \
tags=block:rampup \
cycles=100k \
threads=auto \
--progress console:2s
note

You can also run a command replacing the phase rampup with main-read or main-write to execute other activities defined in the cql-keyvalue workload file.

The threads parameter, specifies the number of concurrent threads used to load the data.

Customise nosqlbench workflows

Nosqlbench uses workflows to define the load activity. You can define your own workflow to satisfy specific loading/benchmarking needs.

Check the workflow details

To check the details of the several predefined workloads and activities, you can dump the definition to a file. To have the list of all the pre-compiled workloads execute:

./nb --list-workloads

The above command will generate the list of pre-compiled workloads like:

# An IOT workload with more optimal settings for DSE
/activities/baselines/cql-iot-dse.yaml

# Time-series data model and access patterns
/activities/baselines/cql-iot.yaml

# A workload with only text keys and text values
/activities/baselines/cql-keyvalue.yaml

To edit a particular workload file locally, you execute the following, replacing the placeholder WORKLOAD_NAME with the name of the workload:

./nb --copy WORKLOAD_NAME

The command generates a file called cql-keyvalue.yaml containing the specifications for the keyvalue workload.

Create your own workload

Workload files can be modified and then executed with nb using the command option workload=WORKLOAD_NAME.

The tool expects the file WORKLOAD_NAME.yaml to be in the same directory of the nb command. If you create the file called my-workload.yaml in the same directory of nb command, the new workload can be run with this command line:

./nb run                   \
driver=cql \
workload=my-workload
tip

You can check the data load using cqlsh as mentioned in the dedicated document.