> This page is for version 10.
> For other versions, use one of these documentation indexes:
> - 11 (default): https://docs.nvidia.com/base-command-manager/product-manuals/11/llms.txt
> - 10: https://docs.nvidia.com/base-command-manager/product-manuals/10/llms.txt

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/base-command-manager/product-manuals/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/base-command-manager/product-manuals/_mcp/server.

# Introduction

Download PDF

## 1.1  Cloud Computing Vs Edge Computing

Cloud computing is traditionally about the concept of end users using resources that are located in a cloud elsewhere. The cloud is the central coordinator, and end users use the resources that are in the cloud rather than using their local resources.

As computing power has become cheaper over time, and resource use has grown, it has in many cases become more financially attractive to shift the emphasis of resource coordination, from the center of the cloud (core of the cloud), over to the local resources which are at the *edge*. These local devices are then called *edge nodes*.

A strong case for edge computing is when the following resource requirements are easier to provide locally via local devices, than via central processing in the cloud:

* low latency

* high bandwidth consumption

* high CPU cycle consumption

For example, a self-driving car requires a low latency, high bandwidth, and high CPU cycle consumption in order to ensure a speedy and safe response to traffic requirements. Attempting to run a self-driving car via central processing in the cloud would be impractically slow or prohibitively dangerous.

Generally, edge computing is regarded as a way to have a geographically spread-out cluster make more local use of its computing resources. A geographically spread-out cluster typically already has plenty of CPU cycles, bandwidth, and low latency at its regular nodes. So, for such a geographically spread-out cluster, making more local use of its computing resources tends to mean granting extra autonomy to the edge computing devices, and making them more independent of the head node.

To achieve this greater autonomy, NVIDIA Base Command Manager uses an *edge director*. This is somewhat similar to the cloud director, but is required to be geographically close to the edge nodes, and is also optimized for edge requirements.

## 1.2  High Speed Monitoring And Local Processing

The importance of being local and autonomous is often due to the environment that the regular nodes are in. The environment is typically under high speed monitoring by many sensors linked to the regular nodes. The data values obtained by the sensors are processed very quickly by the nodes. Such high speed processing of the monitoring data values can typically only reasonably be achieved by the nodes managing the processing locally as much as possible, rather than having the nodes managed by a head node a large distance away.

---

[Download this manual as a PDF](https://docs.nvidia.com/base-command-manager/manuals/10/edge-manual.pdf)