Review the system requirements and environment configuration before installing EVA.
EVA runs on Kubernetes and uses AI models such as Vision Models, LLMs, and VLMs, so a GPU is required.
EVA runs reliably on systems that meet or exceed the minimum specifications.
For production environments, server resources can be adjusted according to detection frequency, workload priority, and concurrent throughput.
EVA App, Vision, and Agent can run together on a single integrated server. Camera connection management, video analysis, situation assessment, and alert generation are handled on the same server.
The following is the minimum recommended specification for 100 cameras.
| Item | Minimum Specification |
|---|---|
| CPU | 32 cores or more |
| RAM | 96 GB or more |
| GPU | RTX PRO 6000 Blackwell |
| Storage | 2 TB SSD |
| OS | Ubuntu 24.04 |
💡 An integrated deployment simplifies initial setup and operations and is efficient for PoCs or small-to-medium environments.
⚠️ Actual requirements may vary depending on camera resolution, number of cameras, detection interval, and other factors.
For large-scale environments, increase the integrated server resources to support the required operating scale.
EVA runs in Kubernetes and requires the following configuration in either a cloud or on-premises environment.
nvidia-device-plugin