The POWDER Manual

2025-09-26 (8165ad6)

← prev up next →

The POWDER Manual

The Powder Team

Powder is a facility for experimenting on the future of wireless networking in a city-scale "living laboratory" environment. It was one of the first two platforms funded by the Platforms for Advanced Wireless Research program at the National Science Foundation (NSF), receiving support from NSF grants CNS-1827940 and CNS-2346555.

Powder is developed, deployed, and operated by the University of Utah, in partnership with the RENEW team at Rice University.

The Powder facility is built on top of Emulab and is run by the Flux Research Group, part of the Kahlert School of Computing at the University of Utah.

1 Powder Overview

1.1 What is Powder

1.2 Powder status

1.3 Roadmap to using Powder

2 Getting Started

3 Powder starter profiles

4.1 Register for an Account

4.1.1 Join an existing project

4.1.2 Create a new project

4.1.3 Setting up SSH access

4.1.4 Setting up X11

5 Powder and Repeatable Research

6 Basic Concepts

6.2 Experiments

6.2.1 Reserving Resources for Experiments

6.2.2 Extending Experiments

6.4 Physical Resources

6.5 Virtual Machines

7 Creating Profiles

7.1 Creating a profile from an existing one

7.1.1 Copying a Profile

7.1.2 Creating the Profile

7.1.3 Updating a profile

7.2 Capturing Disk Images for Profiles

7.3 Repository-Based Profiles

7.3.1 Updating Repository-Based Profiles

7.3.2 Branches and Tags in Repository-Based Profiles

7.4 Sharing Profiles

7.5 Versioned Profiles

8.1 Radio monitoring

8.1.1 Conducted RF

8.1.2 Radiated RF

8.1.3 Massive MIMO

9 Resource Reservations

9.1 What Reservations Guarantee

9.2 How Reservations May Affect You

9.3 Making a Reservation

9.4 Using a Reservation

9.5 Who Shares Access to Reservations

10 Using Powder: From simulation, to emulation, to over-the-air

10.1 Simulated RF Experimentation

10.2 Conducted RF Environment

10.3 Using the Indoor Over-The-Air Lab

10.4 Outdoor Over-The-Air Environment

10.4.1 Step-by-step walkthrough

10.4.2 Some Important Notes

10.5 Next Steps

11 Describing a profile with python and geni-lib

11.1 A single XEN VM node

11.2 A single physical host

11.3 Two XenVM nodes with a link between them

11.4 Two ARM64 servers in a LAN

11.5 A VM with a custom size

11.6 Set a specific IP address on each node

11.7 RF communication

11.8 Specify an operating system and set install and execute scripts

11.9 Profiles with user-specified parameters

11.10 Add storage to a node

11.11 Debugging geni-lib profile scripts

12 Storage Mechanisms

12.1 Overview of Storage Mechanisms

12.2 Node-Local Storage

12.2.1 Specifying Storage in a Profile – Local Datasets

12.2.2 Allocating Storage in a Running Experiment

12.2.3 Persisting Local Data

12.3 Image-backed Datasets

12.4 Remote Datasets

12.5 NFS Shared Filesystems

12.6 Write-back Storage for Fixed and Mobile Endpoints

12.7 High-performance Storage for Streaming Data

12.7.1 Using CephFS in an Experiment

12.7.2 Limitations and Guidelines for CephFS Usage.

12.8 Storage Type Summary (TL;DR)

12.9 Example Storage Profiles

12.9.1 Creating a Node-local Dataset

12.9.2 Creating an Image-backed Dataset from a Node-local Dataset

12.9.3 Using and Updating an Image-backed Dataset

12.9.4 Creating a Remote Dataset

12.9.5 Using a Remote Dataset on a Single Node

12.9.6 Using a Remote Dataset on Multiple Nodes via a Shared Filesystem

12.9.7 Using a Remote Dataset on Multiple Nodes via Clones

13 The Control Network

13.2 Ensuring your experiment nodes use the correct interfaces

13.2.1 Common causes of control network misuse

13.2.2 Empirically verifying interface use

13.3 What to do if you get a Control Network Violation email

13.4 Why is it so important to avoid the control network?

13.5 When can I use the control network?

14 Advanced Topics

14.1 Disk Images

14.3 Public IP Access

14.3.1 Dynamic Public IP Addresses

14.5 Introspection

14.5.1 Client ID

14.5.2 Control MAC

14.5.3 Manifest

14.5.4 Private key

14.5.5 Profile parameters

14.6 User-controlled switches and layer-1 topologies

14.7 Portal API

15 Virtual Machines

15.1.1 Controlling CPU and Memory

15.1.2 Controlling Disk Space

15.1.3 Setting HVM Mode

15.1.4 Dedicated and Shared VMs

16 Automating Experiment Configuration

16.1 Configuration Management: Ansible

16.2 Ansible in Profiles: geni-lib extensions

16.3 Bootstrapping Ansible in Experiments

16.4 Using Ansible in Experiments

16.4.1 Per-experiment Ansible Facts

16.4.2 emulab-common Ansible Roles

16.4.3 Using emulab-common Roles

16.5 Example: OpenZMS profile

16.6 Example: Kubernetes profile

16.7 Example: workflow-manager profile

17 Hardware and Wireless Environments

17.1 Conducted attenuator matrix resources

17.2 Paired Radio workbench resources

17.3 Indoor OTA lab resources

17.4 Rooftop Base-station resources

17.5 Dense-deployment Base-station resources

17.6 Fixed-endpoint resources

17.7 Mobile-endpoint resources

17.8 Near-edge computing resources

17.9 Cloud computing resources

18 Powder basic srsLTE Tutorial

18.1 Objectives

18.2 Prerequisites

18.3 Logging In

18.4 Creating a simple srsLTE experiment

18.5 Exploring Your Experiment

18.5.1 Experiment Status

18.5.2 Profile Instructions

18.5.3 Topology View

18.5.4 List View

18.5.5 Manifest View

18.5.6 Graphs View

18.5.8 Web-based Shell

18.6 Using the srsLTE tools

18.7 Digging deeper

18.8 Terminating the Experiment

18.9 Taking Next Steps

19 Powder OAI Tutorial

19.1 Objectives

19.2 Prerequisites

19.3 Logging In

19.4 Building Your Own OAI Network

19.5 Exploring Your Experiment

19.5.1 Experiment Status

19.5.2 Profile Instructions

19.5.3 Topology View

19.5.4 List View

19.5.5 Manifest View

19.5.6 Graphs View

19.5.8 Web-based Shell

19.6 Starting OAI Services

19.7 Connecting the UE

19.8 In-Depth OAI Profile Documentation

19.9 Terminating the Experiment

19.10 Taking Next Steps

20 Powder OpenStack Tutorial

20.1 Objectives

20.2 Prerequisites

20.3 Logging In

20.4 Building Your Own OpenStack Cloud

20.5 Exploring Your Experiment

20.5.1 Experiment Status

20.5.2 Profile Instructions

20.5.3 Topology View

20.5.4 List View

20.5.5 Manifest View

20.5.6 Graphs View

20.5.8 Web-based Shell

20.5.9 Serial Console

20.6 Bringing up Instances in OpenStack

20.7 Administering OpenStack

20.7.1 Log Into The Control Nodes

20.7.2 Reboot the Compute Node

20.8 Terminating the Experiment

20.9 Taking Next Steps

21 Citing Powder

22 Getting Help

← prev up next →