Architecture Simplified: GCP - Google Compute Engine

Boot Disks

These are block storage volumes attached to the Google Compute Engine instances on Google Cloud Platform (GCP).

This is the root volume where your images is loaded once you boot the VM instance.

There is only one root volume where your operating system is loaded. Though you can add additional disks but those are not root volume.

Your VM instance and persistent disk are located independent to each other. This is why you can attach one persistent disk to a VM instance at a time and once it is detached from that VM instance, you can attach it to another VM instance.
The data which is stored on the persistence disk can still be available on it even after you detach it from the VM instance - if you select that appropriate option.
Persistence disk performance scales automatically. It means you can resize your persistent disk or can add more persistence disks to your VM instance, based on your requirement.
Google Compute Engines manages the distribution of data to ensure redundancy and optimal performance automatically. It does so by distributing persistent disk data across several physical disks.
Persistent disks can be zonal or regional.
There are ~~three~~ four persistent disk types -

These can be zonal or regional
These are backed by standard HDD (Hard Disk Drives)
It supports all Machine Types
These are highly efficient and reliable block storage
Regional Standard persistent disk provides synchronous replication across two zones in a region
This is default disk type attached to VM
Use Case -

These can be zonal or regional
These are backed by SSD (Solid-State Drives)
This is a cost-effective solution to SSD persistent disks where it balances between performance and cost.
These are cost-effective and reliable block storages
Regional Balanced persistent disk provides synchronous replication across two zones in a region
Use Case -

Since this is a cost-effective alternative to SSD PD, so for the same maximum IOPS as SSD PD and lower IOPS per GB.
Balanced PD offers performance levels suitable for most general purpose applications.

These can be zonal or regional
These are backed by SSD (Solid-State Drives)
These are fast and reliable block storage.
Regional Balanced persistent disk provides synchronous replication across two zones in a region
Use Case -

suitable for high-performance database needs and enterprise applications.
This can be used for low latency (for example - single-digit millisecond latency) requirement.
This provides higher IOPS compared to Standard PD.

These are currently available in only few zones.
These can be zonal only
These are backed by SSD (Solid-State Drives)
Unlike other disk types, you can chose your desired IOPS level (in the range of 10K to 120K IOPS)
In order to get maximum performance levels offered by extreme persistent disks, you must attach your extreme persistent disks to virtual machine (VM) instances that are large machine types like M1, or M2, or N2-64 and larger machine types..
Image can not be created from this disk type.
This disk type can not be resized if mounted to a VM instance.
Use Case -

It provides high performance for both random access workloads and bulk throughput.
It is designed for high-end database workloads.

Unlike Persistent disks, Local SSDs are physically attached to the VM instances.
Since these are physically attached to one VM instance, so it is available to single instance only.
These are designed to provide higher IOPS and lower latency compared to PDs.
Once the VM instance is stopped or deleted, the data stored on local SSDs are deleted. So, be sure to store data of permanent value somewhere else.
Local SSD's performance depends on the interface you choose. There are two interfaces -

Learn more on Google Cloud Services