Compute

Introduction to Compute

In the context of cloud and developer services, “Compute” refers to the processing power required to run applications, process data, and execute logic. It is the fundamental building block of any software service.

Types of Compute

Types of Compute: A Comparison

Type	Abstraction Level	Control	Management Overhead	Startup Time	Use Case
Virtual Machines (VMs)	Hardware	High (OS access)	High (Patching, Scaling)	Minutes	Legacy apps, specific OS requirements.
Containers	OS Kernel	Medium (Runtime)	Medium (Orchestration)	Seconds	Microservices, portable apps.
Serverless	Runtime	Low (Code only)	Low (No ops)	Milliseconds	Event-driven, APIs, glue code.
Edge Compute	Network Edge	Low (Code only)	Low (Global dist.)	Milliseconds	Low-latency apps, personalization.

1. Virtual Machines (VMs)

Emulation of a physical computer. Provides full control over the OS and environment.

• Examples: Amazon EC2, Google Compute Engine.

2. Containers

Lightweight, portable units that package code and dependencies. Shares the host OS kernel.

• Examples: Docker, Kubernetes.

3. Serverless

Run code without provisioning or managing servers. The cloud provider handles scaling and infrastructure.

• Examples: AWS Lambda, Cloudflare Workers.

4. Edge Compute

Running processing tasks closer to the user to reduce latency.

• Examples: Cloudflare Workers, Fastly Compute@Edge.

Key Concepts

• CPU vs. GPU: CPU for general-purpose processing; GPU for parallel processing (AI/ML, rendering). CPU vs GPU (Intel).
• Memory (RAM): Temporary storage for active data.
• Throughput: The amount of data processed in a given time.
• Latency: The time it takes to start processing a request.

Recap

• Compute is the processing power for applications.
• VMs offer control but high overhead.
• Containers offer portability.
• Serverless removes infrastructure management.
• Edge Compute minimizes latency by running code near the user.