1.2 Fundamental Cloud Concepts
Key Takeaways
- Cloud shifts six business attributes at once — flexibility, scalability, reliability, elasticity, agility, and TCO — and each maps to a specific business outcome
- Moving from CapEx to OpEx removes stranded capacity, refresh cycles, and hidden facilities costs from the customer's TCO
- Private, hybrid, and multicloud each fit different business scenarios based on isolation, portability, and best-of-breed needs
- Network infrastructure vocabulary — IP, ISP, DNS, regions, zones, fiber optics, subsea cables, edge data centers, latency, bandwidth — is testable on the CDL exam
- Google Cloud's global private fiber backbone, 40+ regions, and 180+ edge POPs are what make scalability, reliability, and low latency real
Six Cloud Concepts That Change Business Outcomes
The exam applies six attributes to business scenarios. Each flips a legacy constraint:
| Concept | Definition | Business example |
|---|---|---|
| Flexibility | Mix and match service types, regions, vendors | A bank runs managed PostgreSQL in production and a graph database in one region for fraud detection |
| Scalability | Capacity grows to meet demand without redesign | A streaming service adds transcoding capacity for a surprise viral event |
| Reliability | Service continues through failure | Multi-zone BigQuery keeps analytics available during a zone outage |
| Elasticity | Capacity also shrinks when demand drops | An e-commerce site releases capacity after the holiday peak, paying only for what it used |
| Agility | Time from idea to production measured in hours | A fintech ships a feature daily behind feature flags |
| Total cost of ownership (TCO) | Full cost of owning and operating a workload | Cloud OpEx replaces hidden costs of data-center staff, power, cooling, and refresh cycles |
Flexibility is breadth of choices; scalability is the ceiling; elasticity is the floor; reliability is uptime; agility is velocity; TCO is the financial lens that ties them all to the chief financial officer (CFO). A common exam trap is conflating scalability with elasticity: scalability is the ability to grow, elasticity is the ability to both grow and shrink on demand. A workload that scales up but never releases capacity is scalable but not elastic, and the TCO difference is large — the elastic workload pays for the trough, the merely scalable one keeps paying for the peak.
CapEx to OpEx Shift and Its Effect on TCO
Traditional IT is capital expenditure (CapEx): servers bought up front, depreciated over 3–5 years, sitting idle during lows. Cloud is operating expenditure (OpEx): metered like electricity — pay for what you use this month, expense it on the income statement.
The shift affects TCO in four ways:
- Removes stranded capacity — no 40% idle headroom bought "just in case."
- Removes refresh cycles — the provider upgrades hardware; you do not depreciate it.
- Adds elasticity savings — release capacity after peak instead of owning it year-round.
- Surfaces hidden costs — power, cooling, real estate, and staff time become the provider's problem; the customer's line item shrinks to usage.
The exam's TCO pitfall: comparing only server sticker price. A real TCO must include facilities, staffing, downtime, opportunity cost of slowness, and refresh cycles — most of which cloud eliminates or reduces.
When Private, Hybrid, or Multicloud Best Applies
| Business scenario | Recommended model | Why |
|---|---|---|
| Regulated hospital, patient data must stay on dedicated hardware you control | Private cloud | Isolation and auditability |
| Startup launching a software-as-a-service (SaaS) minimum viable product with unknown growth | Public cloud | No capex, elastic scale |
| Bank migrating a mainframe-led core to cloud over five years | Hybrid cloud | Sensitive data on-prem now, burst analytics to cloud, portability during transition |
| Global manufacturer acquired by a company standardized on a different provider | Multicloud | Honor both parents' investments; regional data sovereignty |
| European public-sector workload with strict in-country data residency | Hybrid or multicloud | Keep residency-sensitive data local; use public cloud where allowed |
Network Infrastructure Vocabulary
Cloud infrastructure lives on physical networks. The exam tests these terms:
- IP address — a unique number identifying a device or service on a network (IPv4 like
203.0.113.5, IPv6 like2001:db8::1). - Internet Service Provider (ISP) — company that connects end users and organizations to the internet.
- Domain Name System (DNS) — translates human names like
google.comto IP addresses; cloud DNS is itself a managed service. - Regions — independent geographic areas (e.g.,
us-central1,europe-west1) containing one or more zones. - Zones — isolated failure domains within a region; deploying across three zones is the standard for high availability.
- Fiber optics — glass strands carrying light; the high-bandwidth backbone of long-haul networks.
- Subsea cables — fiber laid on ocean floors connecting continents; Google co-owns many (e.g., Equiano, Dunant).
- Network edge data centers / edge POPs — small facilities close to end users that cache content and terminate connections, reducing latency.
- Latency — one-way trip time for a packet; low latency is milliseconds for edge-cached content.
- Bandwidth — data per second a link can carry; high bandwidth serves large files and streams concurrently.
How Google Cloud Supports Transformation With Global Infrastructure
Google's network is its quiet differentiator. Google Cloud runs in 40+ regions and 120+ zones (numbers grow quarterly), connected by Google's private global fiber backbone — not the public internet — with subsea cables Google co-owns. Edge points of presence (POPs) in 180+ cities terminate user connections near them, lowering latency and absorbing distributed denial-of-service (DDoS) attacks. This backbone lets Google Cloud offer services like BigQuery that move terabytes between regions at backbone speed, and lets global apps place replicas close to every major user base. The infrastructure is what makes the six cloud concepts above real: scalability that spans regions, reliability across zones, and low-latency access from almost anywhere.
A retailer's traffic spikes 10x on Black Friday and is steady the rest of the year. Which cloud attribute most directly lowers its total cost of ownership?
A company compares cloud and on-premises TCO using only server sticker prices. Which hidden on-premises cost is most likely to be missed?
Within Google Cloud, what is a "zone"?
A hospital runs patient records on dedicated hardware it controls but wants elastic analytics for research. Which deployment model best fits?