Cloud Deployment Models: Public, Private, Hybrid, Community

Core Concepts: What is a Cloud Deployment Model?

A cloud deployment model defines where and how cloud resources are hosted, deployed, and managed. Unlike service models (SaaS/PaaS/IaaS), which describe what services are delivered, deployment models focus on ownership, access, and infrastructure control.

Why Does It Matter?

Deployment models determine:

• Security (who controls data?)
• Cost (CAPEX vs. OPEX)
• Scalability (how quickly can resources expand?)
• Compliance (does it meet legal/industry standards?)

mindmap
  root((Cloud Deployment Models))
    Public
      "Hosted by 3rd party (AWS, Google Cloud)"
      "Multi-tenant, pay-as-you-go"
      "Examples: eSewa, Daraz, WhatsApp"
    Private
      "Dedicated to single org (NTC, banks)"
      "On-premise or hosted private cloud"
      "High security, full control"
    Hybrid
      "Mix of public + private"
      "Data-sensitive workloads in private, scalable in public"
      "Example: Ncell’s customer data (private) + analytics (public)"
    Community
      "Shared by orgs with common needs (e.g., hospitals)"
      "Cost-sharing, specialized compliance"

1. Public Cloud: Shared Infrastructure, Global Scale

Definition

A public cloud is a third-party-owned infrastructure (e.g., AWS, Microsoft Azure, Google Cloud) where resources (servers, storage, apps) are shared across multiple tenants via the internet. Users pay for what they use (pay-as-you-go).

How It Works

1. Multi-tenancy: A single physical server hosts multiple virtual machines (VMs) for different customers.
2. Elasticity: Resources scale automatically (e.g., Daraz’s Black Friday traffic).
3. Managed by provider: No hardware maintenance for users.

Advantages

Real-World Example: eSewa’s Public Cloud

• Use Case: eSewa uses AWS for its payment gateway and user authentication.
• Why Public?
  • Handles millions of transactions/day without over-provisioning.
  • Cost-effective: Pays only for peak hours (e.g., Dashain/Tihar).
  • Disaster recovery: AWS’s global data centers ensure uptime.
• Challenge: Sensitive financial data must be encrypted (see Unit 5).

WORKED EXAMPLE: Cost Comparison Suppose a startup hosts its web app on:

• Public Cloud (AWS): $0.05/hr for 1 vCPU, 1GB RAM. Runs 24/7 for 30 days. Cost = 0.05 × 24 × 30 = $36/month.
• Private Cloud: $5,000 one-time for a dedicated server + $200/month maintenance. Cost = $5,200 (Year 1) vs. $432 (AWS over 12 months). Insight: Public cloud wins for startups/small businesses; private cloud for long-term, high-security needs.

2. Private Cloud: Dedicated, Secure, and Customizable

Definition

A private cloud is a cloud environment exclusively for a single organization, hosted either on-premise (in their own data center) or by a third party (e.g., IBM Cloud Private).

How It Works

1. Single-tenancy: No shared resources; entire infrastructure is dedicated.
2. Control: Organization manages hardware, software, and security.
3. Customization: Tailored to specific compliance needs (e.g., banking, healthcare).

Advantages/Disadvantages

Real-World Example: NTC’s Private Cloud

• Use Case: Nepal Telecom (NTC) uses a private cloud for:
  • Core network management (SMS, voice routing).
  • Customer data (billing, identity verification).
• Why Private?
  • Regulatory requirements: Telecom data must stay within Nepal (no public cloud).
  • Uptime critical: 99.99% SLA for voice services.
  • Custom integrations: Legacy systems (e.g., old billing software) need private cloud compatibility.

WORKED EXAMPLE: Traffic Routing in Kathmandu Imagine NTC’s private cloud manages SMS traffic during a festival:

1. Peak load: 10 million SMS/day (vs. 1 million normal).
2. Public cloud would scale automatically, but NTC cannot risk latency in emergency alerts.
3. Private cloud solution:
   • Load balancers distribute traffic across dedicated servers.
   • Local caching reduces database load.
   • Result: 99.9% delivery rate vs. 95% if using public cloud.

3. Hybrid Cloud: Best of Both Worlds

Definition

A hybrid cloud combines public and private clouds, with orchestration (software) to enable data/application portability between them.

How It Works

1. Data classification: Sensitive data (e.g., Ncell customer IDs) stays private; analytics (e.g., usage trends) go public.
2. Workload distribution:
   • Private: Databases, APIs, internal tools.
   • Public: Web apps, CDN, batch processing.
3. Orchestration tools: VMware, OpenStack, or cloud-native tools (e.g., AWS Outposts) manage the hybrid setup.

Advantages/Disadvantages

Real-World Example: Ncell’s Hybrid Cloud

• Use Case: Ncell uses:
  • Private cloud: Stores SIM registration data (mandated by Nepal’s law).
  • Public cloud (AWS): Hosts customer app (e.g., Ncell MyAccount) and AI chatbots.
• Data Flow:
  1. User logs into Ncell app (public cloud).
  2. App calls private cloud API to verify identity.
  3. Transaction data (e.g., top-up) is processed in public cloud but linked to private cloud records.
• Benefit: Meets Nepal’s telecom regulations while leveraging AWS’s global reach for app performance.

WORKED EXAMPLE: Pathao’s Hybrid Order Queue Pathao’s ride-hailing system uses a hybrid model:

1. Private cloud:
   • Driver locations (GPS data) stored encrypted.
   • Payment processing (PCI compliance).
2. Public cloud:
   • Matching algorithm (scales during Diwali).
   • User app CDN (faster global load times).
3. Result:
   • Security: Driver data never exposed.
   • Speed: App responds in <1s even with 1M users.

4. Community Cloud: Shared by a Group

Definition

A community cloud is shared by organizations with common concerns (e.g., healthcare, education, government). It can be public, private, or hybrid but is restricted to a specific community.

How It Works

1. Shared infrastructure: Hospitals pool resources for Electronic Health Records (EHR).
2. Custom compliance: Meets healthcare regulations (e.g., Nepal’s Health Data Protection Act).
3. Cost-sharing: Each hospital pays a portion of the total cost.

Advantages/Disadvantages

Real-World Example: Nepal’s Education Cloud

• Use Case: KU and TU share a community cloud for:
  • Student portals (e.g., TU’s result system).
  • Research data repositories.
• Why Community?
  • Avoids duplication: Single instance of student database.
  • Government funding: Shared cost via MoE scholarship portal.
  • Interoperability: KU and TU can exchange transcripts seamlessly.

WORKED EXAMPLE: Kathmandu Traffic Management Imagine a community cloud for Kathmandu Metropolitan City (KMC), NTC, and Pathao:

1. Shared data:
   • Traffic cameras (NTC) + ride demand (Pathao) fed into a single analytics engine.
2. Public cloud: Runs AI traffic prediction (scalable).
3. Private cloud: Stores sensitive probe data (e.g., Pathao’s driver routes).
4. Outcome:
   • Reduced congestion by 15% (as seen in pilot projects).
   • Cost: KMC pays 40%, NTC 30%, Pathao 30%.

Comparison Table: Deployment Models

In the Real World

1. eSewa (Public Cloud)

   • Idea Used: Elastic scalability during festivals.
   • How: AWS auto-scales servers during Dashain (transactions spike 500%).
   • Impact: Avoids crashes and reduces costs vs. over-provisioning.

2. NMB Bank (Private Cloud)

   • Idea Used: Data sovereignty (customer data never leaves Nepal).
   • How: Hosts core banking on on-premise private cloud with blockchain for transactions.
   • Impact: Meets Nepal Rastra Bank (NRB) regulations while preventing breaches.

3. NEPSE (Hybrid Cloud)

   • Idea Used: Public-facing apps + private data.
   • How:
     • Public cloud: Stock trading app (scalable for Diwali rush).
     • Private cloud: Investor KYC data (encrypted, never exposed).
   • Impact: Handles 10K+ trades/day without latency.

4. Kathmandu Hospitals (Community Cloud)

   • Idea Used: Shared EHR system.
   • How: 5 private hospitals share a HIPAA-compliant cloud for patient records.
   • Impact: Doctors access records in <2s (vs. 10s with paper files).

Exam Tip: How to Score Full Marks

1. Define Clearly

   • Start each answer with one-sentence definitions (e.g., "A hybrid cloud is a deployment model combining public and private clouds...").
   • Example: "Private cloud offers single-tenancy, ensuring no resource sharing between tenants."

2. Use Tables for Comparisons

   • Exams often ask to "compare public and private clouds." Always use a table (like above) with 4–5 clear columns.

3. Real-World Examples Are Mandatory

   • Nepal-specific examples (NTC, Ncell, eSewa) fetch extra marks.
   • Structure: "[Company] uses [model] for [purpose] because [reason]."

4. Diagrams > Text

   • Draw a hybrid cloud layered diagram or sequence diagram of data flow (e.g., Ncell’s private→public transition).
   • Mermaid tip: Use stateDiagram-v2 to show "How data moves between private/public clouds."

5. Common Pitfalls to Avoid

   • ❌ Confusing service models (SaaS/PaaS/IaaS) with deployment models.
   • ❌ Saying "public cloud is always secure" (it’s shared, so security depends on provider).
   • ❌ Ignoring cost in comparisons (always mention CAPEX vs. OPEX).

WORKED EXAM ANSWER (5 Marks) Question: "Compare public and private clouds with examples from Nepal. Which would you recommend for a bank and why?"

Answer: Public and private clouds differ in ownership, cost, security, and scalability (see table below). In Nepal:

Recommendation for a Bank: A private cloud is ideal for banks like NMB because:

1. Data sovereignty: Customer records must stay in Nepal (public clouds may host data abroad).
2. Regulatory compliance: Nepal Rastra Bank (NRB) requires audit trails and encryption, which private clouds support.
3. Security: Banks handle sensitive transactions; shared public cloud risks data breaches (e.g., if another tenant is hacked).
4. Legacy systems: Many banks use old COBOL systems that need private cloud integration.

Visual Aid:

sequenceDiagram
    participant Bank as Private Cloud
    participant Customer as Customer App (Public Cloud)
    participant AWS as Public Cloud
    Customer->>AWS: "Login Request"
    AWS->>Bank: "Verify KYC (API call)"
    Bank-->>AWS: "Return: Authenticated"
    AWS-->>Customer: "Grant Access"

Note: The bank’s KYC data stays in private cloud, while the app frontend scales via AWS. This hybrid approach balances security and scalability.