Which of the Following Is a Compute Service?
(And Why That Question Keeps Showing up in Cloud‑Tech Interviews)
Ever stared at a multiple‑choice quiz that asks, “Which of the following is a compute service?The phrasing is vague enough to make you wonder if the answer is “coffee” or “the person who writes code.On top of that, ” You’re not alone. ” and felt the brain‑fart of “wait, what even is a compute service?” In practice, though, the question is a shortcut for testing whether you can separate compute from storage, networking, and management in the world of cloud platforms.
Below is the kind of deep‑dive you’d get if you asked a seasoned cloud engineer to walk you through the whole thing. I’ll break down what a compute service really means, why you should care, how the major providers slice it up, the pitfalls most folks trip over, and finally, a handful of tips you can actually use tomorrow.
What Is a Compute Service
In plain English, a compute service is any cloud offering that lets you run code—anything from a single function to a massive, distributed application. It’s the “brain” of your workload, the part that does the heavy lifting, while storage keeps the data, networking shuffles packets, and management tools keep everything tidy That's the part that actually makes a difference..
Think of a cloud environment as a kitchen. Compute is the stove where you actually cook the meal. Storage is the pantry where you keep the ingredients. Consider this: networking is the hallway that lets the waiter bring the dish to the table. Management is the chef’s notebook with the recipes and timers.
When you hear “compute service,” picture any product that spins up CPU cycles, GPU cores, or server‑less execution environments on demand. The key is that you’re not buying a physical box; you’re renting processing power that scales up or down with your needs.
Core Characteristics
- On‑demand provisioning – spin up an instance in seconds, shut it down when you’re done.
- Scalability – horizontal (more instances) or vertical (bigger instance types).
- Pay‑as‑you‑go – you’re billed for the actual compute time, not for an idle server.
- API‑driven – everything can be scripted, which is why DevOps loves it.
If a service checks those boxes, you’re looking at a compute offering.
Why It Matters / Why People Care
Understanding which services count as compute is more than quiz‑night trivia. It affects cost, architecture, and even security Easy to understand, harder to ignore..
- Cost control – you can’t optimize a bill if you lump compute with storage. A mis‑tagged service could double‑count your spend.
- Architecture decisions – picking a server‑less function versus a full VM changes how you design your app, how you handle latency, and how you manage state.
- Compliance – some regulations require you to know where code runs, especially for workloads that process personal data.
In practice, the wrong assumption can lead you to over‑provision, under‑perform, or even fail an audit.
How It Works (or How to Do It)
Let’s walk through the three big cloud providers—AWS, Azure, and Google Cloud—and see how they label their compute services. I’ll also throw in a couple of niche players for context Which is the point..
Amazon Web Services (AWS)
| Service | Compute? | What It Does |
|---|---|---|
| EC2 (Elastic Compute Cloud) | ✅ | Classic virtual machines. Now, you pick the instance type, OS, and you manage the OS yourself. |
| Lambda | ✅ | Server‑less functions that run in response to events. You only pay for execution time. |
| ECS (Elastic Container Service) | ✅ | Managed container orchestration (Docker) that runs on EC2 or Fargate. |
| EKS (Elastic Kubernetes Service) | ✅ | Managed Kubernetes—still compute, just abstracted behind a k8s API. That said, |
| S3 (Simple Storage Service) | ❌ | Object storage, not compute. Worth adding: |
| RDS (Relational Database Service) | ❌ (mostly) | Managed database; the compute is hidden but you don’t provision it directly. |
| CloudFront | ❌ | CDN for content delivery. |
How to pick: If you need full OS control, go EC2. If you want to avoid servers entirely, Lambda is the go‑to. For container‑first teams, ECS/EKS give you the flexibility of Docker/Kubernetes without the underlying hardware headaches Easy to understand, harder to ignore. Worth knowing..
Microsoft Azure
| Service | Compute? | What It Does |
|---|---|---|
| Virtual Machines | ✅ | Azure’s answer to EC2—full VMs you manage. That said, |
| Blob Storage | ❌ | Object storage, not compute. |
| Azure SQL Database | ❌ (mostly) | Managed relational DB; compute is abstracted. |
| Azure Container Instances (ACI) | ✅ | Run containers without managing VMs or a full orchestrator. |
| Azure Kubernetes Service (AKS) | ✅ | Managed Kubernetes, similar to EKS. Here's the thing — |
| Azure Functions | ✅ | Server‑less, event‑driven code execution. |
| Azure CDN | ❌ | Content delivery, not compute. |
How to pick: Azure mirrors AWS’s lineup, but ACI is a nice middle‑ground if you just need a quick container job without the overhead of a full AKS cluster Easy to understand, harder to ignore..
Google Cloud Platform (GCP)
| Service | Compute? | What It Does |
|---|---|---|
| Compute Engine | ✅ | Google’s VM offering, fully configurable. Which means |
| Google Kubernetes Engine (GKE) | ✅ | Managed Kubernetes. |
| BigQuery | ❌ (mostly) | Data warehouse; the compute is hidden behind SQL queries. Consider this: |
| Cloud Run | ✅ | Fully managed container execution (Knative‑based). |
| Cloud Storage | ❌ | Object storage. |
| Cloud Functions | ✅ | Server‑less functions, similar to Lambda/Functions. |
| Cloud CDN | ❌ | CDN, not compute. |
How to pick: Cloud Run is a hidden gem—run any container, pay per request, and let Google handle scaling. If you already love Docker, it’s often simpler than a full GKE cluster Surprisingly effective..
Niche Players & Edge Cases
- IBM Cloud Functions – server‑less, IBM’s Lambda clone.
- DigitalOcean Droplets – simple VMs, definitely compute.
- Heroku Dynos – abstracted containers; still compute, just “Platform as a Service.”
The pattern is clear: any service that executes code on demand is a compute service, regardless of whether you see the underlying VM.
Common Mistakes / What Most People Get Wrong
-
Confusing managed databases with compute – RDS, Azure SQL, and Cloud SQL are managed services. You don’t spin up the compute yourself, so they’re not considered compute in the typical interview sense.
-
Labeling a CDN as compute – CDNs cache and deliver content, but they don’t run your own code.
-
Thinking “server‑less = no compute” – The word “server‑less” is a marketing term. Under the hood, there’s still CPU cycles, just abstracted away.
-
Mixing up storage‑oriented services that have a compute side – S3 offers event notifications that can trigger Lambda. The notification itself isn’t compute, but the Lambda it triggers is Worth keeping that in mind..
-
Assuming “container service = storage” – Containers need compute to run. Even if a service is marketed as “container registry,” the registry is storage; the runtime (ECS, AKS, Cloud Run) is compute Most people skip this — try not to..
Avoiding these traps shows you actually understand the cloud stack, not just the buzzwords.
Practical Tips / What Actually Works
- Create a quick cheat sheet – list the major services for each provider and mark them ✅/❌. Keep it on your desk for interview prep.
- Use tags in your cloud‑cost reports – tag every compute resource (EC2, VM, Lambda) with “compute” so you can slice spend later.
- Prototype with server‑less first – If you’re unsure whether you need a full VM, spin up a Lambda/Function. It’s cheap, fast, and you’ll instantly see the limits (cold start, execution time).
- take advantage of managed Kubernetes only when you truly need it – Running a k8s cluster just to host a single microservice is overkill; use Cloud Run or ACI instead.
- Watch the “hidden compute” in managed services – Services like RDS still consume CPU; if you’re budgeting, factor in the instance class even if you don’t manage it directly.
FAQ
Q1: Is a managed database like Amazon RDS a compute service?
A: No. RDS is a managed storage service. The underlying compute is hidden, but you don’t provision it yourself, so it’s not counted as a compute service in most contexts Worth keeping that in mind. Turns out it matters..
Q2: Are containers themselves a compute service?
A: The runtime that runs containers (ECS, AKS, Cloud Run) is compute. A container registry (ECR, GCR) is just storage That alone is useful..
Q3: Does a CDN count as compute?
A: Not really. A CDN caches and serves static content; it doesn’t execute your custom code It's one of those things that adds up..
Q4: What about “edge functions” like Cloudflare Workers?
A: Those are server‑less compute at the edge—definitely a compute service.
Q5: How do I know if a new service is compute or not?
A: Ask: “Do I write code that runs on it, and do I pay for execution time?” If yes, you’re looking at compute Simple, but easy to overlook. No workaround needed..
That’s the short version: compute services are the parts of the cloud that actually run your code, whether that code lives in a VM, a container, or a tiny function. Knowing the difference pays off in interviews, on the bill, and when you’re designing the next big thing.
Now go ahead and pick the right compute service for your next project—your future self (and maybe your manager) will thank you.
