Introduction

OpenDT is a Shadow Mode Digital Twin for datacenters. It connects to a real datacenter (or a mock of one) and replays workload data through the OpenDC simulator, comparing predicted power consumption against actual measurements in real time.

Key Capabilities

• Power Prediction: Estimate power consumption based on workload patterns
• What-If Analysis: Test infrastructure changes without touching live hardware
• Active Calibration: Automatically tune simulation parameters to minimize prediction error
• Carbon Estimation: Calculate carbon emissions from power draw and grid intensity

Architecture Overview

OpenDT bridges physical and digital infrastructure through a modular design:

Component	Purpose	OpenDT Implementation
Physical Infrastructure (P)	Data source	dc-mock service
Front-end (B)	User interfaces	Grafana + REST API
Orchestration (C)	Message routing	Kafka
Data Platform (D, E)	Telemetry & storage	Parquet files
Simulator (F-I)	Power prediction	OpenDC + calibrator
Virtual Infrastructure (V)	Digital twin state	Topology configuration

See the Architecture page for detailed component mapping.

Data Flow

1. dc-mock reads historical workload and power data from Parquet files
2. Messages are published to Kafka topics with configurable speed factor
3. simulator consumes workload messages, aggregates them into time windows, and invokes OpenDC
4. api queries results and serves them to the Grafana dashboard

Use Cases

Reproduce Experiments

Run predefined experiments with the Reproducibility Capsule:

make up config=config/experiments/experiment_1.yaml

Explore What-If Scenarios

Modify the datacenter topology and observe how power consumption changes:

• Upgrade CPU architecture
• Change power model parameters
• Add or remove hosts

Validate Power Models

Compare different power model types (MSE, Asymptotic, Linear) against real measurements to find the best fit for your hardware.

Quick Start

See the Getting Started guide to run your first simulation.