Day 1: Introduction to GCP and Cloud Computing

This chapter introduces the fundamental concepts of cloud computing, examining why modern software engineering transitioned away from physical on-premises data centers to cloud platforms. It covers the market landscape comparing Google Cloud Platform (GCP) with AWS and Azure, walks through creating and managing a free-tier GCP account safely, introduces Google Cloud Shell, and synthesizes key cloud architectural vocabulary.

1. The Evolution of Infrastructure: From Laptops to On-Premises Data Centers

To understand cloud computing, consider the legacy lifecycle of application development and deployment.

Real-Life Analogy: The "Flip Zone" E-Commerce Startup (circa 2010)

Imagine an e-commerce startup named Flip Zone building an online store in 2010.

1. Local Development: Developers build the application on their local machines (e.g., 4 CPU cores, 4 GB RAM) and verify that it runs locally.
2. Quality Assurance (QA): The application is handed over to QE/QA engineers, who run it on their own laptops and approve it for production.
3. The Deployment Dilemma: Flip Zone cannot run a commercial platform from a developer or QA laptop for two main reasons:
   • Single Point of Failure: A personal laptop can shut down, disconnect, or crash, leading to complete service downtime.
   • Resource Bottlenecks: A standard laptop cannot handle thousands of concurrent requests without running out of CPU and memory.

To serve public traffic reliably, Flip Zone had to purchase physical enterprise servers from vendors such as IBM or HP. Unlike laptops, physical enterprise servers offer extended compute configurations (e.g., 100 CPUs, 100 GB RAM, 10 TB storage).

To operate these physical servers, companies had to build dedicated server rooms or full-scale Data Centers, complete with organized server racks, networking switches, ethernet cabling, and dedicated System Administration (SysAdmin) teams.

2. Operational & Financial Challenges of On-Premises Data Centers

Managing on-premises infrastructure presents two major bottlenecks: operational overhead and financial over-provisioning.

Challenge A: Operational & Maintenance Overhead

Building an application represents less than half of the total engineering effort; setting up and maintaining the data center demands the majority of continuous resources. SysAdmins were required to manage:

• Environmental Controls: Maintaining strict temperature and humidity controls 24/7 to prevent hardware overheating.
• Operating System Provisioning: Installing and configuring OS layers (Linux, Windows Server) so application software could interface with underlying hardware.
• Security Patching: Upgrading OS versions continuously to eliminate vulnerabilities.
• Hardware Maintenance: Repairing failed physical disks, cabling, switches, and coordinating with hardware vendors (IBM/HP) for replacements.

Conceptually: An Operating System (OS) serves as the indispensable software medium bridging the physical hardware layer and the application software layer. Without OS maintenance, hardware resources remain inaccessible and insecure.

Challenge B: Financial Over-Provisioning & Underutilization

When purchasing physical servers, organizations must provision hardware for maximum peak capacity rather than average usage.

Peak vs. Off-Peak Compute Usage Math

Consider an application's utilization across a 24-hour cycle:

• At 1:00 AM, traffic is minimal. System resource utilization is at $\approx 10\%$.
• At 6:00 AM, traffic rises slightly. System resource utilization reaches $\approx 20\%$.
• At 10:00 AM (Peak Hours), traffic spikes. System resource utilization reaches $\approx 80\%$.

At off-peak hours (e.g., 1:00 AM), $90\%$ of the paid-for compute power sits idle. Because physical hardware requires upfront capital expenditures (CapEx), companies pay $100\%$ of the server procurement and electricity cost regardless of actual traffic.

3. The Birth of Cloud Computing & Pay-As-You-Go Architecture

To solve these industry-wide challenges, cloud providers (pioneered by Amazon with AWS, followed by Microsoft Azure and Google Cloud Platform) built global enterprise data centers.

Multi-Tenant Shared Infrastructure

Instead of every individual business maintaining its own server room, cloud providers purchase physical servers at massive scale (tens of thousands of servers per facility). They then carve out small, isolated slices of these servers for individual companies.

Pay-As-You-Go (Utility Pricing) Model

Cloud computing delivers computing services—including compute power, storage, database management, and networking—over the internet on demand.

Companies no longer buy physical servers upfront. Instead, they rent dynamic compute units and pay strictly for the time and resources consumed:

• At 1:00 AM, Flip Zone requests a small configuration ($2\text{ vCPU}, 2\text{ GB RAM}$) $\rightarrow$ Charged minimal rate $\char36 X$.
• At 6:00 AM, Flip Zone scales up to ($10\text{ vCPU}, 10\text{ GB RAM}$) $\rightarrow$ Charged moderate rate $\char36 Y$.
• At 10:00 AM, Flip Zone scales to peak capacity ($60\text{ vCPU}, 60\text{ GB RAM}$) $\rightarrow$ Charged peak rate $\char36 Z$.

4. Cloud Deployment Models: Public, Private, and Hybrid Cloud

Organizations deploy cloud resources using three primary architectural patterns depending on regulatory, cost, and security requirements.

1. Public Cloud

• Infrastructure is owned, maintained, and operated off-site by third-party cloud providers (GCP, AWS, Azure).
• Multi-tenant model: Multiple organizations share physical hardware while remaining logically isolated.
• Ideal for startups, digital native companies, and public-facing enterprise applications (used by $>90\%$ of modern startups).

2. Private Cloud (On-Premises Cloud)

• Infrastructure is built and dedicated exclusively to a single organization.
• Common in financial institutions, banking systems, and healthcare providers with extreme regulatory, compliance, or strict data-sovereignty mandates.
• High operational cost: The company must maintain dedicated SysAdmin teams and hardware racks.

3. Hybrid Cloud

• Combines private cloud (on-premises) with public cloud infrastructure.
• Data Split Pattern: Sensitive core workloads (e.g., user financial databases, transaction queues) remain inside the private cloud, while non-sensitive workloads (e.g., static web frontends, public documentation, internal dev environments) run on the public cloud.

Cloud Deployment Comparison Matrix

5. Google Cloud Platform (GCP) vs. AWS & Azure

While Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP) offer functional parity across basic compute, storage, and networking services, GCP differentiates itself through scalability, network performance, and foundational infrastructure.

GCP’s Unique Selling Proposition (USP): Proven Massive Scale

Google Cloud is built on the exact underlying infrastructure, global network fibers, and orchestration systems that power Google’s flagship consumer products, including YouTube and Gmail.

• Battle-Tested Scalability: Since Google already handles massive global concurrent throughput for its own applications, GCP offers proven scalability for enterprise workloads.
• Implementation & Innovation Focus: Competition between AWS, Azure, and GCP centers on security, low latency, global throughput, and managed services (e.g., Kubernetes via GKE).

6. Setting Up and Managing a Free Tier GCP Account

Account Hierarchy Differences: Personal vs. Enterprise

• Personal Google Account: Used for hands-on learning. Resources are scoped directly inside Projects.
• Google Workspace (Enterprise): Used in enterprise environments. Establishes a top-level Organization root, under which Folders, Projects, and Resources are structured hierarchically for strict Identity and Access Management (IAM).

Step-by-Step Free Account Provisioning

1. Navigate to cloud.google.com.
2. Click Get started for free using a standard Google account.
3. Complete the registration form:
   • Select Country/Region.
   • Select Account Type (Individual).
   • Enter Payment Information (GCP supports Credit/Debit cards and alternative regional options like UPI in India).
4. Credit Allocation: New users receive $300 USD in free credits (or local currency equivalent, e.g., ₹25,000 INR) valid for up to 90 days.

Free Tier Limits & Avoiding Unexpected Billing

GCP provides a "Always Free" allocation, but usage exceeding these precise limits deducts from the $300 credit pool or incurs charges:

• Compute Engine: 1 free non-preemptible e2-micro VM instance per month (1 vCPU, 1 GB RAM). Creating a second e2-micro instance will incur hourly charges.
• Cloud Storage: 5 GB-months of free Standard Storage.
• BigQuery: 1 TB of querying per month and 10 GB of storage per month.

Note on Payment Security: In regions where bank standing instructions or automatic recurring charges are registered upon payment verification, developers can manually cancel standing instructions via their bank interface to prevent unexpected automated overdrafts after the free trial concludes.

7. Accessing GCP: Web Console vs. Cloud Shell & GCloud CLI

GCP resources can be accessed via three interfaces:

1. GCP Web Console

Accessible directly via browser at console.cloud.google.com. Allows graphical navigation to spin up virtual machines, manage databases, and review billing metrics.

2. Google Cloud Shell

A browser-accessible, hosted terminal environment running an ephemeral Linux virtual machine pre-configured with the gcloud Command Line Interface (CLI), git, docker, and core development utilities.

Key Advantages of Cloud Shell:

• Eliminates the need to install gcloud SDK tools on local developer laptops.
• Automatically authenticates using active GCP console credentials.
• Bypasses local network restrictions, firewalls, and local installation conflicts.

8. Essential Cloud Computing Vocabulary & Mental Models

Mastering cloud engineering requires a working understanding of core terminology:

1. Virtualization & Virtual Machines (VMs)

Virtualization is the foundational technology that abstracts physical hardware into software representation. A software layer called a Hypervisor sits on physical servers (e.g., a host server with 100 CPUs, 100 GB RAM) and slices them into isolated Virtual Machines (VMs) running independent operating systems.

2. Application Programming Interface (API)

An API is a structured program interface that enables software systems to communicate. When a user creates a VM via the Cloud Shell or Console, the request is sent as an HTTP/REST API request to GCP's Compute Engine API infrastructure.

3. Regions & Availability Zones

• Region: A specific geographical location (e.g., us-east1, asia-south1) divided into isolated locations called Availability Zones.
• Availability Zone (AZ): A distinct physical data center with independent power, cooling, and networking within a Region.
  • Why multi-AZ deployment matters: Deploying an application across multiple AZs ensures that if one physical data center suffers a power failure or natural disaster, traffic automatically fails over to an operational data center nearby.

4. Scalability, Elasticity, & Agility

• Scalability: The capacity of an application or system infrastructure to handle increasing workload volume (e.g., handling 10,000 to 1,000,000 concurrent requests).
• Elasticity: The system's ability to dynamically add resources during traffic spikes and automatically decommission them when demand drops to reduce costs.
• Agility: The speed at which engineering teams can provision new resources, test ideas, and deploy software updates without waiting weeks for hardware procurement.

5. High Availability, Fault Tolerance, & Disaster Recovery

• High Availability (HA): Architectural design ensuring systems remain operational and accessible without downtime ($99.99\%$ uptime targets).
• Fault Tolerance: The system's ability to continue operating uninterrupted even when individual underlying hardware or software components fail completely.
• Disaster Recovery (DR): Business continuity strategies, tools, and automated data backups that restore operations following catastrophic failures or complete regional data center outages.

6. Load Balancing

A mechanism that distributes incoming network traffic across multiple server instances or regions to prevent any single resource from becoming overwhelmed.

Core Cloud Vocabulary Reference

Summary and Key Takeaways

• On-Premises Drawbacks: Traditional server rooms require capital investment (CapEx) and operational overhead (OS updates, temperature control, cabling), while forcing companies to over-provision hardware that sits idle during off-peak hours.
• Cloud Computing Shift: Cloud computing turns computing infrastructure into an utility-style, pay-as-you-go resource, replacing fixed CapEx costs with dynamic operational expenditure (OpEx).
• Deployment Models: Choose Public Cloud for agility and elastic scaling, Private Cloud for dedicated on-premises infrastructure under strict regulatory constraints, or Hybrid Cloud to combine on-premises database security with public cloud elasticity.
• GCP Enterprise Advantage: Google Cloud Platform leverages the global infrastructure, low-latency network routes, and hardware optimization that powers billion-user services like YouTube and Gmail.
• Safe Account Management: New GCP accounts include $300 USD in free credits over 90 days. Avoid charges by staying within free-tier limits (such as a single e2-micro instance per month).
• Cloud Shell Utility: Google Cloud Shell provides an in-browser Linux machine pre-authenticated with the gcloud CLI, removing local SDK installation overhead.
• Hierarchy Awareness: Personal Google accounts organize resources within Projects, whereas enterprise accounts use Google Workspace to manage resources under an Organization $\rightarrow$ Folder $\rightarrow$ Project $\rightarrow$ Resource structure.
• Foundational Architectural Principles: Scalability, elasticity, high availability, fault tolerance, and load balancing form the core architecture needed to build resilient enterprise cloud applications.