If a VM runs for more than 25% of a month, GCP automatically starts lowering the hourly rate. If it runs 24/7 for the whole month, you get up to a 30% discount automatically.

• Best for: Casual or unpredictable workloads where you forgot to plan, but the VM ended up running for a long time anyway.

2. Committed Use Discounts (CUDs)

You officially promise GCP that you will use a certain amount of vCPUs/RAM (Resource-based) or spend a specific dollar amount (Spend-based) for 1 or 3 years. In return, GCP cuts your bill by up to 70%.

• Best for: Predictable, permanent, 24/7 production workloads that you know your company will run for years.

3. Spot VMs

GCP sells you their leftover, unused computer power at a massive 60% to 91% discount. The catch? GCP can shut down your VM with only a 30-second warning. If a regular customer needs that hardware.

• Best for: Jobs that can be safely interrupted and restarted later—like data processing, batch jobs, testing environments, or stateless apps.

What is Observability?

If Monitoring tells you that something is broken, Observability tells you why it broke.

In GCP, observability is handled by a suite of tools called the Google Cloud Operations Suite. It relies on three main pillars: Metrics, Logs, and Traces.

The Three Pillars of GCP Observability

1. Metrics (Cloud Monitoring) — The Dashboard Gauges

What it is: Numeric data measured over time. It tells you the health and performance of your system.

In GCP: It shows you graphs of CPU usage, memory usage, network traffic, or disk read/write speeds on your VMs or Kubernetes clusters.

2. Logs (Cloud Logging) — The Black Box Recorder

What it is: A text-based, chronological diary of every single event that happens in your system.

In GCP: It records things like "User Tony logged in at 11:25 PM", "Database connection failed", or "Auth service returned a 500 error."

3. Traces (Cloud Trace) — The Delivery Tracker

What it is: Tracks a single user request as it travels through a complex system of multiple microservices or VMs

In GCP: If a user clicks "Buy Now" and the webpage takes 5 seconds to load, Cloud Trace shows you: Frontend took 0.1s → Auth service took 0.2s → Payment Gateway took 4.7s. You instantly know where the bottleneck is.

What are sole-tenant nodes?

Imagine renting the entire physical building for yourself. Nobody else can live there. You can put your own family members (your individual VMs) into different rooms, but absolutely no strangers are allowed inside the physical structure.

Why would someone pay extra for this?

• Security or compliance: PCI-DSS, HIPAA, FedRAMP.
• Licensing: some software, like Windows Server or SQL Server, is licensed per physical core or processor.
• Company policy: requires dedicated hardware.

A sole-tenant node is an entire physical server dedicated to you. It works with the same machine types and images.

What is Live Migration in GCP?

Google sometimes needs to do maintenance on the physical host your VM runs on. With live migration, your running VM is moved to another host in the same zone without stopping and without changing its properties. Triggers include infrastructure maintenance and hardware failures.

You configure it under the availability policy:

1. On host maintenance: Migrate (the default when supported) moves the VM to other hardware, while Terminate stops it.
2. Automatic restart: restarts the VM if it was terminated for non-user reasons, such as a hardware failure.

1. Custom Image

A Custom Image is a snapshot of a boot disk containing an operating system (OS) along with any pre-installed software, configurations, or data.

• How it works: Instead of starting from a plain, vanilla OS image provided by Google (like standard Ubuntu or Debian), you take an existing VM disk, install your application dependencies, security patches, or software stacks, and create a custom image from it.

Key Benefits:

• Faster Boot Times: Because applications and packages are already baked into the image, your VMs don't need to download or install them from scratch every time they start.
• Consistency ("Golden Images"): Ensures every server in your environment starts with the same software version and security configuration.

Note: You must have the VM in GCP before creating a custom image.

Step 1: Prepare Your Source VM

1. Go to the VM instances page in the GCP Console.
2. Identify the VM whose disk you want to capture as an image.
3. Stop the VM. (Important: To ensure data consistency and avoid corrupting files, the instance should be stopped before creating an image from its boot disk).

Step 2: Navigate to the Images Page

1. In the Google Cloud Console search bar at the top, type Images and select the Images (Compute Engine) service.
2. At the top of the page, click + Create image.

Step 3: Configure Your Custom Image

• Name: Give your image a lowercase, hyphen-separated name (e.g., my-custom-app-image-v1).
• Source: Select Disk.
• Source disk: Choose the stopped VM instance or its specific boot disk from the dropdown menu.
• Location: Choose whether you want the image stored in a Regional location (tied to a specific region for faster local creation) or Multi-regional (available across multiple regions).
• Encryption: (Optional) Leave as Google-managed encryption key unless your organization requires a Customer-Managed Encryption Key (CMEK).
• Family: (Optional) Enter an image family name (e.g., web-servers). Image families point to the latest image in that group automatically when used in templates.
• Guest OS features: GCP usually auto-detects features like UEFI or multi-IP support based on your source disk, but you can configure them manually if needed.

Step 4: Create the Image

1. Click Create at the bottom of the page.
2. GCP will take a few minutes to process the snapshot and build the image. Once complete, it will appear in your list of custom images, ready to be used in Instance Templates or new VM creations.

2. Startup Script (Bootstrapping)

A Startup Script is a script (typically Bash for Linux or PowerShell for Windows) that you pass to a VM instance via metadata. The VM executes it automatically during boot.

How it works: When the VM boots up, the Compute Engine guest agent picks up the startup script from the instance metadata and runs it with root/administrator privileges.

How to add a script while creating a VM with a specific software configuration.

• Start creating your VM instance or Instance Template with all the configuration, and in Advance option, add the script.

Bootstrapping startup script works!!

3. Instance Template

An Instance Template is a global resource that acts as a reusable blueprint for creating VM instances or Managed Instance Groups (MIGs).

• How it works: It packages all the configuration settings required to launch a VM—such as the machine type (e.g., e2-medium), the boot disk (which can reference a Custom Image), network settings, metadata (including Startup Scripts), and IAM service accounts.
• Key Benefits:
  • Reusability: Instead of manually configuring dozens of flags every time you create a VM, you define it once in a template.
  • Auto-scaling: Instance templates are required for Managed Instance Groups (MIGs) to add or remove identical instances based on traffic load automatically.

Here is the step-by-step guide to creating one:

Step 1: Navigate to Instance Templates

1. In the Google Cloud Console search bar at the top, type Instance templates and select it under Compute Engine.
2. At the top of the page, click + Create instance template.

Step 2: Configure the Template Settings

An instance template looks almost identical to the normal VM creation page. Fill out the configuration you want your future VMs to use:

• Name: Give your template a lowercase, hyphen-separated name (e.g., my-web-server-template).
• Machine configuration: Choose your series, machine type (e.g., e2-medium), and memory.
• Boot disk: Click Change if you want to swap out the default OS. This is where you can select your Custom Image from the dropdown menu instead of a public image.
• Firewall: (Optional) Check the boxes to allow HTTP/HTTPS traffic if this is a web server.

Step 3: Add Your Startup Script (Bootstrapping)

1. Scroll down and click on Management, security, disks, networking, sole tenancy to expand it.
2. Click on the Management tab.
3. Scroll down to the Automation section.
4. Paste your script directly into the Startup script text box (or provide your Cloud Storage URL in the metadata section if preferred).

Create a Single VM Instance from a Template and click on create.

VM is ready.

The automated script is also working!!

Once you have created an Instance Template, you can use it to quickly spin up individual virtual machines or automate scaling for your applications.

Hands-on: On IP Address

Set Up a VM with an Ephemeral IP

Goal: Show how a standard dynamic IP address changes when a VM is stopped and restarted.

1. Go to the GCP Console and navigate to Compute Engine > VM instances.
2. Click Create Instance.
3. Keep the default settings (e.g., e2-medium, Debian/Ubuntu OS).

4. Expand Networking > click on the default Network interface

5. Under External IPv4 address, ensure Ephemeral is selected.

Note down the External IP address listed next to your new VM:

6. Click the SSH button next to the VM to connect to it.
7. Inside the terminal, run curl ifconfig.me to verify that the VM sees the same external IP.

8. Go back to the GCP Console, select the VM, and click Stop. Wait a minute, then click Start.

• Observe: The VM now has a completely different External IP address. Explain to your audience that ephemeral IPs are released back into Google's pool when a resource is stopped.

Start it again, and you will see the external IP again.

Promote to a Static External IP

Goal: Show how to lock down an IP address so it survives restarts.

1. In the GCP Console search bar, search for VPC network and select IP addresses.

2. You will see the current ephemeral IP assigned to your running VM.

3. Click the three dots (More actions) next to your VM's IP address and select Promote to static IP address

4. Give it a name (e.g., demo-static-ip) and click Reserve.

• Go back to Compute Engine > VM instances.
• Select your VM and click Stop.
• Once stopped, click Start.
• Observe: The External IP address remains the same: 136.114.228.233

Clean Up (Important)

Goal: Prevent unnecessary cloud costs. GCP charges for static external IP addresses that are reserved but not attached to a running resource.

1. Delete the VM instance.
2. Go back to VPC network > IP addresses.
3. Select your static IP (which should now show as "In use by: None").
4. Click Release Static IP.

Takeaway: ephemeral IP addresses change whenever a GCP resource stops and restarts, while static IP addresses remain permanently locked to your project until you manually delete them.

• Ephemeral IPs: Free while the VM is running, but they are released back into Google's pool when the VM stops.
• Static IPs: Reserved exclusively for you. They survive restarts and deletions, but GCP charges a small fee if they sit idle.

Google Compute Engine lets you create a virtual computer in the cloud. Google manages the physical server, and you use a virtual machine (VM) to run your applications.

A VM has resources such as:

CPU: How much processing work it can do.
Memory (RAM): How much data and how many programs it can handle at once.
Disk: Where the operating system, files, and applications are stored.
Network: How it communicates with other systems and the internet.

For example, suppose you create a Linux VM to practice commands, install Docker, and run a small application. You need to decide how powerful that virtual computer should be.

Key building blocks

What is a machine family?

A machine family is a category of VMs designed for a particular kind of workload.

Analogy: Think of buying a vehicle. You might choose a small car, a truck, or a sports car depending on your needs. They are all vehicles, but each is designed for different work.

GCP machine families work similarly.

Remember: The family tells you the general purpose and design of the VM. It does not tell you the exact amount of CPU and RAM you will get.

What is a machine type?

A machine type tells you the actual CPU and RAM resources your VM will receive.

For example, imagine you choose the E2 machine series. You can select a smaller or larger machine type depending on your workload.

Here is an example using E2 machine types:

Let's break down one example: e2-standard-2 --> 2 vCPUs, 8 GB RAM
e2 = the machine series.
standard = a balanced CPU-to-memory configuration.
2 = two vCPUs.

This is a general-purpose VM configuration suitable for many small applications and learning exercises.

Hands-on: How to choose a machine family and type in GCP and create VMs.

Step 1 — Open Compute Engine

Go to the Google Cloud Console, select your project, then navigate to Compute Engine → VM instances → Create instance.

Step 2 — Select a machine configuration

Find the Machine configuration section. Depending on the current console layout, you'll see options for the machine family, series, and machine type.

Step 3 — Choose a family and type

For a basic learning VM, select General-purpose → E2, then choose an appropriately small machine type available to your project and zone.

Step 4 — Choose OS and storage

*An image decides what OS and software your VM runs.
*Public images are ready-made (Debian, Ubuntu, Windows Server, SQL Server).

Step 5 — Choose networking for communication.

Step 6 — Review and create

Hands-on: HTTP server

Connect to the VM via SSH

Task: Install Apache Server on VM

Your takeaway

Remember this simple rule: Choose the family based on the workload, then choose the machine type based on the resources you need.