---
title: Workloads
slug: workloads
docTags: 
createdAt: 2026-04-20T16:24:42.637Z
---

Workloads are Amberflo’s way of representing **internal systems or components** that you want to attribute usage and cost to.

They provide a structured way to answer a simple question:

**Where is this usage happening, and what is it costing us?**

## What is a Workload?

A workload represents a **unit of infrastructure or functionality** within your system.

Examples include:

- An image generation pipeline
- A customer support chatbot
- A recommendation engine
- A background processing job
- GPU clusters or compute resources
- Any service or system that drives usage

Workloads are not tied to customers. They are **internal constructs** used to organize and analyze usage and cost.

## How Workloads Fit into the System

Workloads are used during **metering**.

When you send usage data, you can attach a workload identifier. That usage is then:

- **Attributed to that workload**
- **Rated into cost using your cost models**
- **Aggregated into spend views**

The same usage data can also be used for billing. You are not forced to separate them.

You can:

- Use metering **only for cost tracking**
- Use metering **only for billing**
- Use the same meter for **both cost and billing**

Workloads sit on the cost side, but they work with the same underlying usage data.

## Workloads Page Overview

Navigate to the **Workloads** page to view and manage your workloads.

### What You’ll See

Each row in the table represents a workload and includes:

- **Workload Name**
- **Current Spend (Month-to-Date)**
- **Created Date**

### Expanding a Workload

Click on a workload row to expand it.

You will see:

- A list of **previous months**
- Spend totals for each month

### Drill Into Monthly Spend

Click on a specific month to view detailed cost data.

This opens a breakdown similar to an invoice, where you can see:

- All usage attributed to that workload
- How that usage was rated into cost
- The components contributing to total spend

This is how you move from high-level totals to detailed cost analysis.

## Creating a Workload

### Navigate

- Go to the **Workloads** page
- Click **Create Workload** (top right)

### Required Fields

A workload consists of two fields:

- **Name**
  A human-readable label
  Example: Image Generation Pipeline
- **Workload ID**
  A unique identifier used in meteringConstraints:
  - Letters, numbers, hyphens, underscores only
  - No spaces

Click **Save** once complete.

## How Workloads Are Used in Metering

The **Workload ID** is critical.

When you send usage data to a meter, you include this ID as part of the workload dimension.

Example flow:

1. Create a workload with ID image\_pipeline
2. Send usage with workload = image\_pipeline
3. Amberflo attributes that usage to the workload
4. Cost is calculated and aggregated under that workload

Without this linkage, usage cannot be attributed to internal systems.

## Why Workloads Matter

Workloads allow you to:

- Understand **which parts of your system drive cost**
- Compare spend across different services or pipelines
- Identify **high-cost components**
- Track trends over time
- Make informed optimization decisions

This is typically used by **engineering and infrastructure teams** to:

- Analyze system efficiency
- Monitor cost growth
- Prioritize optimization efforts

## Summary

Workloads are the foundation for internal cost attribution.

They:

- Represent parts of your infrastructure
- Receive attributed usage through metering
- Aggregate cost based on your rate models
- Enable detailed analysis through spend views

If you want to understand how your system consumes resources and where your costs come from, workloads are where that starts.