ale93p / namb

Not only A Micro-Benchmark

namb

NAMB (Not only A Micro-Benchmark) is a parametrizable and automatically generated benchmark application for Data Stream Processing Systems (DSP).

Features

• Simple YAML-based configuration for parameters definition

• Automatic application generation, no need to manually touch the platform APIs

• Ready-to-use python script to deploy the benchmark application

• Supported platforms:

  platform	version
  Apache Flink
  Apache Storm
  Apache Heron

Requirements

• Java 8 and maven environment
• Python 3

Quick Start

N.B. We assume a local deployment of a DSP platform is previously configured and running.
For this quickstart we consider running NAMB on Apache Flink.

1. Clone the repository in a local folder and move in it

2. Compile the project:

   python3 namb.py build

   or directly with maven:

   mvn clean install

3. Prepare NAMB environment

   • Copy the default NAMB configuration file to the main configuration folder:

   cp conf/defaults/workflow_schema.yml conf/namb.yml

   • Copy the platform-specific configurations to the main configuratino folder:

   cp conf/defaults/flink-benchmark.yml conf/flink-benchmark.yml

   More info on how to ustomize the configuration files see doc/configurations

4. Run the benchmark application:

   python3 namb.py flink

   Use -h for more options and infos.

Publications

[pagliari2019workflow] A. Pagliari, F. Huet, G. Urvoy-Keller, "Towards a High-Level Description for Generating Stream Processing Benchmark Applications", IEEE BigData, 3rd IEEE BPOD Workshop, 2019

[pagliari2020namb] A. Pagliari, F. Huet, G. Urvoy-Keller, "NAMB: A Quick and Flexible Stream Processing Application Prototype Generator", IEEE/ACM CCGrid 2020