Benchmark in Docker Environment

Pre-built Docker Images

We provide two base Docker images quay.io/assetopsbench/assetopsbench-basic and quay.io/assetopsbench/assetopsbench-extra which serves as the AssetOpsBench environment. The base images provide

• quay.io/assetopsbench/assetopsbench-basic: A minimal Python 3.12 environment with only essential Python packages installed. Suitable for lightweight benchmarking tasks or custom setup.
• quay.io/assetopsbench/assetopsbench-extra: A full-featured Python 3.12 environment with all components pre-installed, including Individual agents (e.g., IoT Agent), MetaAgent, and AgentHive.

The python environment in both docker images use Conda to manage. The name of pre-built Conda environment is assetopsbench. For full list of installed packages, please refer to basic_requirements.txt and extra_requirements.txt.

CouchDB Setup

To streamline access to asset data during benchmarking, we use a Dockerized CouchDB instance. This setup allows for easy loading, querying, and management of benchmark datasets within a consistent environment.

A sample docker-compose.yml is provided to help you start CouchDB locally with default configuration and credentials. We use couchdb_setup.sh as starting command to create a CouchDB database populating with a sample Chiller dataset.

Run Benchmark Script

In docker-compose.yml, we mount the benchmark scripts (entrypoint.sh and run_plan_execute.py and scenario files as volumes to assetopsbench container.

Now we run the following command,

cd /path/to/AssetOpsBench
chmod +x benchmark/entrypoint.sh
docker-compose -f benchmark/docker-compose.yml build
docker-compose -f benchmark/docker-compose.yml up

You can use .env or add your environment variables to .env which can be passed to containers through docker-compose.yml.

Note that in entrypoint.sh, we activate the python environment by conda activate assetopsbench first.

Customization

Customize Your Python Environment

You can install additional packages (e.g. your own agents) by using quay.io/assetopsbench/assetopsbench-basic or quay.io/assetopsbench/assetopsbench-extra as base image. To build your own image, here is an example Dockerfile,

FROM quay.io/assetopsbench/assetopsbench-basic

ARG CONDA_ENV_NAME=assetopsbench
COPY your_requirements.txt /tmp/your_requirements.txt
RUN opt/conda/bin/conda run -n ${CONDA_ENV_NAME} pip install -r /tmp/your_requirements.txt

Customize Your Run Script

Use your own run script and update docker-compose.yml. Recommend mount your scripts as volumes for ease development.