Machine types and families (E2, N2, C2) in GCP - Deep Dive

Overview - Machine types and families (E2, N2, C2)

What is it?

Machine types in Google Cloud Platform (GCP) are predefined sets of virtual hardware resources like CPU and memory. Families like E2, N2, and C2 group machines optimized for different tasks. E2 machines are cost-effective and flexible, N2 machines balance performance and price, and C2 machines focus on high compute power. These options help users pick the right virtual machine for their needs.

Why it matters

Choosing the right machine type saves money and improves application performance. Without these options, users might pay too much or get slow results. For example, running a heavy calculation on a weak machine wastes time, while using a powerful machine for simple tasks wastes money. Machine families solve this by matching resources to workload needs.

Where it fits

Before learning this, you should understand basic cloud computing and virtual machines. After this, you can learn about custom machine types, autoscaling, and workload optimization in GCP.

Mental Model

Core Idea

Machine types are like different car models designed for specific driving needs, balancing power, efficiency, and cost.

Think of it like...

Imagine choosing a car: a small sedan (E2) for daily errands, a mid-size SUV (N2) for family trips, and a sports car (C2) for speed. Each fits different purposes and budgets, just like machine types in the cloud.

┌───────────────┐   ┌───────────────┐   ┌───────────────┐
│    E2 Family  │   │    N2 Family  │   │    C2 Family  │
│ Cost-effective│   │ Balanced perf.│   │ High compute  │
│ Flexible CPUs │   │ More memory   │   │ Fast CPUs     │
└──────┬────────┘   └──────┬────────┘   └──────┬────────┘
       │                   │                   │
       ▼                   ▼                   ▼
  Suitable for         Good for general     Best for CPU-
  lightweight tasks    workloads needing    intensive tasks
                       moderate power       like gaming or
                                            scientific work

Build-Up - 7 Steps

1

Foundation

What is a machine type in GCP

Concept: Introduces the basic idea of machine types as virtual hardware profiles.

In Google Cloud, a machine type defines how many CPUs and how much memory a virtual machine (VM) has. Think of it as choosing the size and power of a computer you rent in the cloud. Each machine type comes with fixed resources you can use to run your applications.

Result

You understand that machine types control the computing power and memory available to your cloud VM.

Knowing machine types is key to matching your cloud computer's power to your application's needs.

2

Foundation

Understanding machine families

Concept: Explains that machine types are grouped into families based on their design and purpose.

Machine families group similar machine types that share characteristics. For example, E2 machines focus on cost savings, N2 machines balance performance and price, and C2 machines offer high CPU power. This grouping helps users quickly find machines suited for their workload.

Result

You can identify that machine families help organize machine types by their strengths and typical uses.

Recognizing families simplifies choosing the right machine by focusing on workload goals.

3

Intermediate

E2 family: cost-effective and flexible

🤔Before reading on: do you think E2 machines prioritize performance or cost savings? Commit to your answer.

Concept: Details the E2 family as budget-friendly machines with flexible CPU options.

E2 machines are designed to be affordable and flexible. They use shared CPU resources and can burst performance when needed. They are great for small to medium workloads that don't need constant high power, like web servers or development environments.

Result

You know E2 machines save money by sharing CPUs and are good for variable workloads.

Understanding E2's flexibility helps avoid overpaying for steady high performance when it's not needed.

4

Intermediate

N2 family: balanced performance and price

🤔Before reading on: do you think N2 machines offer more memory or more CPU power compared to E2? Commit to your answer.

Concept: Introduces N2 machines as a middle ground with better performance and memory than E2.

N2 machines provide a balance between cost and performance. They use newer CPUs and offer more memory per CPU than E2. This makes them suitable for general-purpose workloads like databases, batch processing, or medium traffic web apps.

Result

You understand N2 machines fit workloads needing steady, moderate power without high cost.

Knowing N2's balance helps pick machines that avoid both underpowering and overspending.

5

Intermediate

C2 family: high compute power

🤔Before reading on: do you think C2 machines are best for memory-heavy or CPU-heavy tasks? Commit to your answer.

Concept: Explains C2 machines as optimized for CPU-intensive workloads.

C2 machines have the fastest CPUs and highest CPU-to-memory ratio. They are ideal for tasks needing raw compute power like scientific simulations, gaming servers, or video encoding. They cost more but deliver top performance.

Result

You recognize C2 machines as the choice for demanding CPU workloads.

Understanding C2's focus prevents choosing underpowered machines for heavy compute tasks.

6

Advanced

Choosing machine types for workloads

🤔Before reading on: would you pick E2, N2, or C2 for a web app with fluctuating traffic? Commit to your answer.

Concept: Shows how to match machine families to real workload needs.

For light, variable workloads, E2 machines save money with flexible CPU use. For steady, moderate workloads, N2 machines provide reliable performance. For heavy compute tasks, C2 machines deliver speed. Consider workload patterns, cost, and performance needs when choosing.

Result

You can select machine types that optimize cost and performance for your specific workload.

Knowing workload characteristics guides efficient machine type selection, avoiding waste.

7

Expert

Performance trade-offs and hidden costs

🤔Before reading on: do you think higher CPU count always means better performance? Commit to your answer.

Concept: Explores subtle performance factors and cost implications beyond CPU and memory specs.

More CPUs don't always mean better performance due to shared resources, CPU architecture, and network limits. E2 machines share CPUs which can cause variability. N2 and C2 use dedicated CPUs but cost more. Also, memory bandwidth and disk speed affect real-world results. Understanding these trade-offs helps optimize both cost and performance.

Result

You grasp that machine specs are starting points; real performance depends on many factors.

Recognizing hidden performance factors prevents costly surprises and improves workload tuning.

Under the Hood

GCP machine types define virtual CPUs (vCPUs) and memory allocated to a VM. E2 machines use shared physical CPUs with dynamic allocation, allowing cost savings but variable performance. N2 and C2 machines use dedicated physical CPUs with newer architectures for consistent performance. The hypervisor manages resource allocation, balancing isolation and efficiency.

Why designed this way?

Google designed these families to meet diverse customer needs: E2 for cost-sensitive users, N2 for balanced workloads, and C2 for high-performance computing. Shared CPUs in E2 reduce costs but can cause variability, while dedicated CPUs in N2 and C2 ensure performance at higher prices. This tiered approach balances price and power.

┌───────────────┐       ┌───────────────┐       ┌───────────────┐
│   User VM     │       │   User VM     │       │   User VM     │
│  (E2 Family)  │       │  (N2 Family)  │       │  (C2 Family)  │
└──────┬────────┘       └──────┬────────┘       └──────┬────────┘
       │                       │                       │
       ▼                       ▼                       ▼
┌───────────────┐       ┌───────────────┐       ┌───────────────┐
│ Shared CPUs   │       │ Dedicated CPUs│       │ Dedicated CPUs│
│ (variable)    │       │ (newer gen)   │       │ (highest perf)│
└──────┬────────┘       └──────┬────────┘       └──────┬────────┘
       │                       │                       │
       ▼                       ▼                       ▼
┌───────────────┐       ┌───────────────┐       ┌───────────────┐
│ Hypervisor    │       │ Hypervisor    │       │ Hypervisor    │
│ manages CPU   │       │ manages CPU   │       │ manages CPU   │
│ allocation    │       │ allocation    │       │ allocation    │
└───────────────┘       └───────────────┘       └───────────────┘

Myth Busters - 4 Common Misconceptions

Quick: Do you think E2 machines always provide consistent CPU performance? Commit to yes or no.

Common Belief: E2 machines give the same CPU power all the time like other machine types.

Reality: E2 machines use shared CPUs that can vary in performance depending on other users' workloads.

Why it matters: Assuming consistent performance can cause unexpected slowdowns in applications relying on steady CPU power.

Quick: Do you think more CPUs always mean faster application performance? Commit to yes or no.

Common Belief: Adding more CPUs always speeds up your application linearly.

Reality: Some applications don't scale well with more CPUs due to software limits or resource contention.

Why it matters: Overspending on CPUs without performance gains wastes money and resources.

Quick: Do you think N2 machines are always better than E2 because they have newer CPUs? Commit to yes or no.

Common Belief: N2 machines are always the best choice because they have newer CPUs than E2.

Reality: E2 machines can be more cost-effective for workloads that don't need constant high performance, despite older CPUs.

Why it matters: Choosing N2 unnecessarily increases costs without proportional benefits for some workloads.

Quick: Do you think C2 machines have more memory than N2 machines? Commit to yes or no.

Common Belief: C2 machines always have more memory than N2 machines because they are higher performance.

Reality: C2 machines have higher CPU-to-memory ratios, meaning less memory per CPU compared to N2.

Why it matters: Using C2 for memory-heavy workloads can cause performance issues due to insufficient memory.

Expert Zone

1

E2 machines' shared CPU model can cause noisy neighbor effects, where other users' workloads impact your VM's performance unpredictably.

2

N2 machines support custom machine types allowing fine-tuning of CPU and memory beyond predefined sizes for workload optimization.

3

C2 machines use the latest Intel Cascade Lake CPUs with AVX-512 instructions, beneficial for specialized compute tasks like machine learning.

When NOT to use

Avoid E2 machines for latency-sensitive or high-performance workloads; use N2 or C2 instead. For memory-intensive applications, prefer N2 or custom machine types with more memory. If cost is not a concern but maximum CPU performance is, choose C2. Alternatives include custom machine types or sole-tenant nodes for dedicated hardware.

Production Patterns

In production, teams often use E2 for development and testing to save costs, N2 for general application servers and databases, and C2 for batch processing or scientific computing. Autoscaling groups mix machine types to balance cost and performance dynamically. Monitoring CPU and memory usage guides switching machine types as workloads evolve.

Connections

Autoscaling in Cloud

Builds-on

Understanding machine types helps configure autoscaling policies that pick the right VM sizes automatically based on workload demand.

Cost Optimization

Same pattern

Choosing the right machine type is a key cost optimization strategy, similar to picking the right tool size in manufacturing to avoid waste.

Automobile Engine Types

Analogy from a different field

Just like choosing between a fuel-efficient engine or a high-performance engine affects car behavior and cost, selecting machine families affects cloud VM performance and price.

Common Pitfalls

#1 Choosing E2 machines for high-performance computing needs.

Wrong approach:

gcloud compute instances create my-vm --machine-type=e2-medium

Correct approach:

gcloud compute instances create my-vm --machine-type=c2-standard-4

Root cause:

Misunderstanding that E2 machines share CPUs and are not designed for sustained high CPU workloads.

#2 Assuming more CPUs always improve performance without checking application scaling.

Wrong approach:

gcloud compute instances create my-vm --machine-type=n2-standard-16

Correct approach:

gcloud compute instances create my-vm --machine-type=n2-standard-4

Root cause:

Not analyzing if the application can use multiple CPUs effectively, leading to unnecessary cost.

#3 Using C2 machines for memory-heavy workloads without enough RAM.

Wrong approach:

gcloud compute instances create my-vm --machine-type=c2-standard-8

Correct approach:

gcloud compute instances create my-vm --machine-type=n2-highmem-8

Root cause:

Ignoring that C2 machines have less memory per CPU, causing memory bottlenecks.

Key Takeaways

Machine types in GCP define the CPU and memory resources of virtual machines, grouped into families like E2, N2, and C2.

E2 machines are cost-effective and flexible but share CPUs, causing variable performance suitable for light workloads.

N2 machines balance performance and price with dedicated CPUs, fitting general-purpose workloads.

C2 machines offer the highest CPU performance with dedicated resources, ideal for compute-intensive tasks but with less memory per CPU.

Choosing the right machine type based on workload needs prevents wasted cost and ensures efficient performance.