General requirements

The Milestone AI Bridge is designed to run in a containerized environment using Ubuntu Linux. However, you can run Milestone AI Bridge on any other AMD64 (x86_64) / ARM64 -based Linux system.

Cloud vs. on-premise environments

Although the Milestone AI Bridge architecture is prepared for cloud environments, the current combination of configuration options and software capabilities mainly targets on-premise solutions that are not necessarily fully open to internet-based cloud solutions.

Processing server hardware

Milestone AI Bridge runs on linux-based docker images, as do the majority of IVA applications, and must therefore be run on machines with a Linux operating system.

You can run Milestone AI Bridge on any other AMD64 (x86_64) / ARM64 -based Linux system and Milestone AI Bridge has specifically been tested to run on the NVIDIA Jetson Xavier and Orin devices.

Any machine that fulfills the Ubuntu Server 22.04.1 Long-term support (LTS)  hardware requirements can be used as a processing server, but as is usual with large, data-based, computational analysis, the better the server specifications, the faster and more reliable the results.

Milestone AI Bridge also utilizes one or more NVIDIA GPU cards for video analysis and any machines dedicated to running Intelligent Video Analysis must be correctly configured with one or more NVIDIA GPU cards.

Processing server software

The following must be considered when planning the installation and deployment of your processing server: 

Linux distribution

While any Linux distribution can in theory be used, it is recommended to use Ubuntu Server 22.04.1 Long Term Service (LTS) or later. Milestone recommends using Ubuntu 22.04.1 LTS server and this documentation assumes the use of a Ubuntu 22.04.1 LTS server or later.

Hosting Linux on a Hyper-V virtual machine

If you want to run the Processing server on a Hyper-V virtual machine hosted by a Windows Server operating system, the Discrete Device Assignment feature is required. This feature is available on Windows Server 2019 or later.

Virtualization and GPU passthrough must be enabled in the BIOS of the host machine and the following BIOS settings must be enabled on the Windows server host machine:

  • Intel VT for Directed I/O (VT-D)

  • Trusted Execution Technology (TXT)

  • ASPM (Active State Power Management) or PCI Express Native Power Management (to increase system performance)

  • Single root I/O virtualization (SR-IOV)

Secure Boot may prevent NVIDIA GPU drivers from running and it is recommend you disable Secure Boot in the BIOS all host machines.

XProtect VMS

Milestone AI Bridge is compatible with XProtect 2022 R1 and later. XProtect VMS versions 2022 R1 and 2022 R2 require that the administrators of the XProtect VMS install version-specific patches.

See Milestone AI Bridge support matrix

Containers

The processing server must be able to run Linux Containers. This is typically obtained by installing a Linux server with Docker Compose or Kubernetes.

Docker Compose

You can deploy Milestone AI Bridge on any system supporting Docker Compose on Ubuntu Linux, or any other Linux distribution. If you select to deploy Milestone AI Bridge on a non-Ubuntu Linux distribution, it is advised to thoroughly test the system prior adding using the system for normal operations.

Docker Compose is usually used for a single processing server.

While Milestone AI Bridge is compatible with all versions of Docker Compose version, Milestone AI Bridge has been tested using Docker version 27.3.1.

Kubernetes

When using Kubernetes, a helm chart can be employed for faster installation and deployment of Milestone AI Bridge and IVA applications. You can use Kubernetes to deploy Milestone AI Bridge on any Linux machine that meets the processing server requirements.

Kubernetes can be used for managing multiple processing servers.

While Milestone AI Bridge is compatible with all versions of Kubernetes, Milestone AI Bridge has been tested using Kubernetes version 1.31.1.