1. Home
  2. Cloud Platforms
  3. Getting Started with Intel + OnLogic AI Inference Server for AWS Kinesis Video Streams

Getting Started with Intel + OnLogic AI Inference Server for AWS Kinesis Video Streams

[1] Document Revision History

Document VersionDateChanges
1.011.18.24Initial Release

[2] Solution Overview

The OnLogic Axial Edge Server for KVS is designed to simplify the challenges of video streaming for edge to cloud inference applications. By using OnLogic’s 1U Axial Edge Server powered by Intel 13th Gen Core i Processors and Intel Arc discrete GPU, you can leverage edge processing to supercharge your vision applications. The OnLogic Axial Edge Server for KVS is tested for compatibility with KVS to easily integrate into existing KVS workflows.

Solution Reference Architecture

[3] Hardware Description

[3.1] Hardware Information

A full range of resources are available for your OnLogic Axial Edge Server for KVS on its’ product page under the download section.

AC101 Datasheet and Dimensional Drawings

AC101 Product Manual

AC101 UEFI Manual

AC101 BMC Manual

AC101 Technical Support and Resources

[3.2] Standard Package Contents

Each OnLogic AC101 includes the computer itself, depending on chosen accessories the following may be included. Accessories may be purchased separately on onlogic.com where stated.

  • Mounting hardware for rack deployments or wall mounting hardware.
  • C13 to AC Outlet Power Cables (Selection available for US/EU/UK)

While the configuration is selectable by the user, the reference platform tested contained the following specifications. System requirements will depend on # of streams, local storage requirements, stream specifications (frame rate/resolution/etc.)

CPU: i7-13700E | GPU: Intel Arc Pro A60 | RAM: 16GB | Storage 128GB | Ubuntu 22.04 LTS

[3.3] User Provided Items

OnLogic Axial Edge Server for KVS is designed to integrate with existing and new video streaming applications and are only a part of the solution. The following items are required to take advantage of the solution:

  • IP Capable Video Cameras (RTSP)
  • Edge networking equipment capable of providing video streams to the AC101. The AC101 is equipped with 10GbE and 1GbE LAN. For multi-stream deployments a network switch is recommended.
  • Users may opt to use integrated storage on AC101 if chosen for Offline video storage, however users may also want to consider on-site NAS if consuming large amounts of data. Full specifications on in-server storage options can be found on the AC101 product page.
  • User provided edge AI inference application. The user should be able to optimize using the Intel OpenVINO toolkit to take full advantage of hardware.
  • Edge to Cloud networking to AWS Cloud. The reference architecture for the solution recommends a 5G Modem, but hardwired connectivity is also sufficient.
  • User maintained AWS EC2 instances for KVS and storage platform of their choice such as S3 Buckets.

[4]Set Up Your Deployment Environment

[4.1] File Downloads

The following uses files available from the OnLogic Support site version of this guide . Please use the available download link, this also includes sample video files (.mp4) for ease of use. The solution can run either on CPU or GPU. Files are included for both.

[4.2] Host Setup

  1. On CPU host, install Docker: https://docs.docker.com/engine/install/ubuntu/ .
    1. Perform post-install steps so you do not need to run Docker with `sudo ` https://docs.docker.com/engine/install/linux-postinstall/
  2. Copy *Dockerfile*, *build.sh*, *run.sh*, *C0012_1080p25.mp4*  to CPU host 
  3. execute “build.sh” to start building the docker container
  4. When Docker build is complete, open up the “run.sh”, enter AWS credentials values for environmental variables `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY` and `AWS_DEFAULT_REGION` fields. 
  5. Update the `-w` to your desired working directory. This will be the directory the Docker container will be in when launched
  6. Enter the Docker container by executing `run.sh`
  7. CPU ONLY: Launch the demo with the command `/opt/demo/kvsWebrtcClientMasterGstSample Intel-DLStreamer-Ch` where `kvsWebrtcClientMasterGstSample` is the AWS KVS WebRTC C SDK demo application and `Intel-DLStreamer-Ch` is the WebRTC Signaling channel name.
  8. Launch a web brower on the local host or remote platform and enter the following address: https://awslabs.github.io/amazon-kinesis-video-streams-webrtc-sdk-js/examples/index.html. Make sure the enter the correct AWS region, Access key and Secret key. The channel name will be `Intel-DLStreamer-Ch`.

[5] Setup Edge Server

[5.1] Device Setup

For in-depth support documentation on the OnLogic AC101, you can visit the support page:

  1. Your system will come equipped with either a single or redundant power supply along with AC power cables. You will need to plug either a single or both (redundant only) AC power cables into the system in order to power on the system.
  2. In order to set up the system you’ll need a keyboard, mouse, and display in order to perform the first time setup. Start the system by pressing the power button (if you selected auto-power on your system will start when plugged in).
  3. Install Ubuntu 22.04 LTS or your operating system of choice (If you have chose not to have OnLogic pre-install it)
  4. Follow the instructions outlined in section [4]

[5.2] Deployment Best Practices

  • Ensure you are using a PC that is rated for your environment: operating temperature, input voltage, shock, vibration
  • Ensure to adhere to any certification requirements for IT equipment: Class 1 Division 2, UL Requirements, etc.
  • Ensure systems are installed properly following instruction materials when installing into racks or secured using wall mount.
  • Ensure installation technicians are properly trained in setting up devices on site
  • Prototype before field installation of larger deployments

Users that have appropriate expertise should expect device setup to take less than 30 minutes for your device to be ready to connect to your AWS KVS instance depending on network availability.

As the solution does not include any production environments in AWS cloud as such the user is responsible for setup, deployment, and maintenance of their AWS applications and following AWS best practices to ensure operational excellence and security. OnLogic Axial Edge Server for KVS can be used in any AWS Region or Availability Zone that supports KVS. Please ensure to adhere to any importation requirements if deploying outside the United States or European Union as there may be additional regulatory requirements for IT equipment.

[6] About AWS Kinesis Video Streams

To learn more about AWS Kinesis Video Streams, see Kinesis Video Streams: How it Works and Getting started with Amazon Kinesis Video Streams.

[7] Troubleshooting/Support

OnLogic does not offer technical support for software issues related to AWS services. Contact your technology integration resources or partner to help resolve software issues. Due to the number of different systems OnLogic offers we have a support resource site to answer commonly asked questions and offer preliminary troubleshooting information.

For Hardware support you can contact the OnLogic support team at support@onlogic.com or 802-861-2300 ext. 2 (8:30 AM to 5:00 PM EST). For responses via email and voicemail we strive to respond the same business day. If received after normal business hours we will respond in the order that they were received. Once an RMA is received normal turnaround when parts are available is 5 business days

All OnLogic systems come with a standard 2 year warranty (unless otherwise stated). For additional information the following resources are available:

Updated on November 18, 2024

Was this article helpful?

Keep In Touch.
Subscribe today and we’ll send the latest product and content updates right to your inbox.
Subscribe