Compute Engine Boot Disks for the PCA Exam: Type, Size, and OS

Boot disks are one of those Compute Engine settings that look like trivia until you actually have to architect a workload. The Professional Cloud Architect exam expects you to know what a boot disk is, what choices you make when you create one, and how it differs from the other storage you can attach to a VM. This article walks through the three configuration decisions that matter (operating system, type, and size) and then closes with the boot disk versus attached disk distinction that the exam likes to test.

What a boot disk actually is

Every Compute Engine VM needs a boot disk. The boot disk holds the operating system and the files the VM needs to start up. Without it, there is nothing to boot.

One detail that matters for the exam: boot disks must always be persistent disks. Persistent disks are block storage that is durable and lives independently of the VM lifecycle, so the data on them survives even when the VM is stopped or deleted. That property is what makes a persistent disk suitable as a boot disk in the first place.

Operating system

The first decision when you create a boot disk is which operating system goes on it. The OS determines the software environment the VM runs, which is why this choice is workload-driven rather than a default you accept without thinking.

Compute Engine offers a long list of options. You do not need to memorize the catalog for the Professional Cloud Architect exam. What you do need is the shape of the decision:

• For most general-purpose workloads, a standard Linux distribution like Debian or Ubuntu is fine.
• For container workloads, Container-Optimized OS is the focused choice.
• For machine learning workloads, Deep Learning on Linux ships with the frameworks and drivers preinstalled.

The pattern is general OS for general work, specialized OS for specialized work.

Boot disk type

The second decision is the type of persistent disk that backs the boot disk. There are four to know:

• Balanced. The general-purpose option. A reasonable mix of performance and cost, suitable for most workloads.
• Extreme. The maximum-performance option, with cost to match. Designed for applications with very high I/O demands, like databases that need consistent low latency at high throughput.
• SSD. Fast performance with strong IOPS. Good for high-performance applications that need speed without paying for the absolute peak.
• Standard. The cheapest option. Built on spinning disks rather than flash, which is fine for workloads where I/O is not the bottleneck.

The way to think about it on the exam: if the workload is a database or anything I/O-bound, you are choosing between SSD and Extreme. If it is a general application server, Balanced is the safe answer. If the question emphasizes cost and tolerates lower performance, Standard.

Boot disk size

The third decision is size. The default boot disk size is 10 GB, which is enough to hold the OS and basic system files. You can resize it at any time, so the initial value is not a permanent commitment.

You would increase the size when the VM needs more room for system files, application binaries, or logs. The signal that you need more capacity is disk utilization climbing above the 70 to 80 percent range. That is the threshold to remember if a question gives you a utilization metric and asks what to do.

Two properties of resizing are worth knowing:

• Resizing does not change the disk type. If you started with a Balanced disk and grew it from 10 GB to 200 GB, it stays Balanced. The performance characteristics carry over.
• Cost depends on both size and type. A larger Extreme disk costs significantly more than a larger Standard disk, so you balance the capacity you actually need against the type you actually need.

Boot disk versus attached disk

Once you understand boot disks, the next thing the Professional Cloud Architect exam wants you to distinguish is the boot disk from an attached disk. They are both persistent disks in many cases, but they play different roles.

The boot disk:

• Contains the operating system and the files needed to boot the VM.
• Can be resized, but is typically reserved for the OS rather than used as a data store.

An attached disk:

• Provides extra storage for data or applications, separate from the boot disk.
• Can be added or removed independently of the boot disk and the VM lifecycle.
• Some types persist even if the VM itself is deleted, which is what lets data outlive the compute that produced it.

The architectural reason this separation exists is modularity. The OS lives on one disk, the application data lives on another, and you can scale, resize, replace, or detach the data storage without touching the boot disk. That makes the VM easier to manage, easier to scale, and easier to recover. If the VM dies, an attached persistent disk holding the application data can be reattached to a new VM and the workload continues.

On the exam, when you see a question about where the operating system lives versus where the application data should live, the answer is almost always to keep them on separate disks for the modularity reason above. Attached disks are the answer for application data, scratch space, and anything you want to outlive a single VM.

What to remember

Three configuration decisions: OS (workload-driven), type (Balanced for general, SSD or Extreme for I/O-heavy, Standard for cost), and size (default 10 GB, resize when utilization passes 70 to 80 percent, type does not change on resize). One conceptual distinction: the boot disk holds the OS and the attached disk holds the data, and they are managed independently.

My Professional Cloud Architect course covers Compute Engine boot disks alongside the rest of the compute material.