Cloudflare Containers Demo

Cloudflare Containers let you run code written in any programming language, built for any runtime, as part of apps built on Workers. You can read more about the roadmap on the Cloudflare Blog.

With Containers you can run:

• Resource-intensive applications that require CPU cores running in parallel, large amounts of memory or disk space

• Applications and libraries that require a full filesystem, specific runtime, or Linux-like environment

• Existing applications and tools that have been distributed as container images

Sample App

Checkout the sample Titanic Survival Prediction App to see how we are running an ML model in a Cloudflare Container on the edge.

Architecture

The following diagram shows the request flow from client to container, and how the worker maintains containers through scheduled jobs:

%%{init: {'theme':'base', 'themeVariables': {'primaryColor': '#ffffff', 'primaryTextColor': '#666666', 'primaryBorderColor': '#999999', 'lineColor': '#666666', 'arrowheadColor': '#666666', 'edgeLabelBackground': '#ffffff', 'clusterBkg': '#ffffff', 'background': '#ffffff', 'mainBkg': '#ffffff', 'secondBkg': '#ffffff', 'tertiaryBkg': '#ffffff'}}}%%
graph LR
    Client["🌐 Client"]
    Worker["⚡ Cloudflare Worker<br/>(Hono App)"]
    DO["🏠 Durable Object<br/>(FastAPIContainer)"]
    Container["🐳 Container<br/>(FastAPI App)"]

    %% Request Flow (1-4)
    Client -->|"1️⃣ HTTP Request<br/>/api/*"| Worker
    Worker -->|"2️⃣ getContainer()<br/>Forward Request"| DO
    DO -->|"3️⃣ container.fetch()<br/>Proxy to Container"| Container
    
    %% Response Flow (4-6)
    Container -.->|"4️⃣ FastAPI Response<br/>JSON/HTML"| DO
    DO -.->|"5️⃣ Proxy Response<br/>Back to Worker"| Worker
    Worker -.->|"6️⃣ HTTP Response<br/>To Client"| Client

    %% Scheduled Container Management (Bottom)
    Scheduler["⏰ Cron Scheduler<br/>(Every 2 minutes)"]
    
    Scheduler ==>|"🔄 Trigger Cron Job"| Worker
    Worker ==>|"🚀 container.start()<br/>Pre-warm Container"| DO
    DO ==>|"⚡ Start/Keep Alive"| Container

    %% Styling
    classDef clientStyle fill:#e1f5fe,stroke:#01579b,stroke-width:2px
    classDef workerStyle fill:#f3e5f5,stroke:#4a148c,stroke-width:2px
    classDef doStyle fill:#e8f5e8,stroke:#1b5e20,stroke-width:2px
    classDef containerStyle fill:#fff3e0,stroke:#e65100,stroke-width:2px
    classDef schedulerStyle fill:#fce4ec,stroke:#880e4f,stroke-width:2px

    class Client clientStyle
    class Worker workerStyle
    class DO doStyle
    class Container containerStyle
    class Scheduler schedulerStyle

Loading

Request Flow

1. Client sends HTTP requests to /api/* endpoints
2. Cloudflare Worker (Hono app) receives and routes the requests
3. Durable Object (FastAPIContainer) acts as a container manager and proxy
4. Container (FastAPI app) processes the actual business logic
5. Responses flow back through the same chain

Scheduled Container Management

• A Cron Scheduler runs every 2 minutes to keep containers warm
• It triggers the Worker to pre-warm containers via the Durable Object
• This ensures containers are ready to handle incoming requests quickly

Future Roadmap

Containers has some exciting future features announced on their roadmap.

• Higher limits and larger instances

  • Current limits: 40 GiB memory and 40 vCPU total. Select customers already scale to thousands of containers. Limits will be raised in coming months for larger workloads.

• Global autoscaling and latency-aware routing

  • Future: Global autoscaling and routing to nearest container instances. Enable with a single line of code for automatic load balancing.

class MyBackend extends Container {
  defaultPort = 8080;
  autoscale = true; // global autoscaling on - new instances spin up when memory or CPU utilization is high
}

// routes requests to the nearest ready container and load balance globally
async fetch(request, env) {
  return getContainer(env.MY_BACKEND).fetch(request);
}

• More ways to communicate between Containers and Workers

  • Enhanced Worker-Container communication: exec commands, HTTP handlers, bindings to Cloudflare services via Volumes and programmatic setup.

Pricing & Instance Types

Available Instance Types

Name	Memory	CPU	Disk
dev	256 MiB	1/16 vCPU	2 GB
basic	1 GiB	1/4 vCPU	4 GB
standard	4 GiB	1/2 vCPU	4 GB

Billing Model

Pay-per-use billing: charges start when containers receive requests or are manually started, stop when containers sleep after timeout. Enables easy scale-to-zero with high utilization for bursty traffic.

Compute Pricing (per 10ms)

Resource	Rate	Workers Standard Included
Memory	$0.0000025 per GiB-second	25 GiB-hours
CPU	$0.000020 per vCPU-second	375 vCPU-minutes
Disk	$0.00000007 per GB-second	200 GB-hours

Egress Pricing

Region	Rate per GB	Workers Standard Included
North America & Europe	$0.025	1 TB
Australia, New Zealand, Taiwan, Korea	$0.050	500 GB
Everywhere else	$0.040	500 GB