System Maintenance & CASE Tools: Types, Phases, Tools & Economic Feasibility

TAKEAWAYS:

• System maintenance is not just fixing bugs but also adapting to new laws (e.g., NEPSE’s updated trading rules) and improving performance (e.g., Daraz’s order queue optimization).
• CASE tools automate SDLC phases: Upper CASE (requirements → design) and Lower CASE (coding → testing) reduce errors by 30–50% (used by banks for loan processing systems).
• Economic feasibility uses payback period, NPV, and ROI—eSewa’s online payment system had a 3-year payback despite high initial costs.
• Documentation must follow SMART rules (Simple, Meaningful, Accurate, Relevant, Timely) to avoid system failures like Kathmandu’s traffic management software glitches.
• Implementation methods (direct, parallel, phased) are chosen based on risk tolerance—NTC uses phased rollout for its new billing system to minimize downtime.
• CASE tools like ERwin (ER diagrams) and Rational Rose (UML) cut design time by 40%—used by Pathao for ride-matching algorithms.

1. System Maintenance: Types, Importance, and Real-World Impact

System maintenance ensures a system remains functional, efficient, and aligned with business goals after implementation. It is not optional—poor maintenance leads to system failures (e.g., NEPSE’s 2023 trading halt due to outdated software).

Types of System Maintenance

Use this state diagram to visualize the maintenance cycle:

stateDiagram-v2
    [*] --> Corrective: Bug fixes (e.g., eSewa login crashes)
    Corrective --> [*]
    [*] --> Adaptive: Legal/tech changes (e.g., Ncell’s new telecom regulations)
    Adaptive --> [*]
    [*] --> Perfective: Performance upgrades (e.g., Daraz’s faster checkout)
    Perfective --> [*]
    [*] --> Preventive: Security patches (e.g., banks’ fraud detection updates)
    Preventive --> [*]

Why it matters:

• eSewa’s 2022 outage cost Rs. 50M in lost transactions—corrective maintenance was reactive and costly.
• Ncell’s adaptive maintenance for 5G required updating its billing system to handle higher data speeds.

2. CASE Tools: Automating SDLC with Upper/Lower CASE

CASE (Computer-Aided Software Engineering) tools reduce human error and speed up development. They are classified into:

Types of CASE Tools

mindmap
  root((CASE Tools))
    Upper CASE
      Requirements Analysis: CORE, RequisitePro
      Design: ERwin (ERD), Rational Rose (UML)
    Lower CASE
      Coding: Visual Studio, Eclipse
      Testing: Selenium, JUnit
    I-CASE (Integrated)
      Full SDLC: IBM Rational, Microsoft Visio + TFS

Real-World Example:

• Pathao’s ride-matching algorithm uses Lower CASE tools (Python + Selenium) for automated testing to ensure no driver-passenger mismatches.
• NTC’s billing system uses Upper CASE (ERwin) to model customer-service interactions before coding.

3. Economic Feasibility: Payback Period, NPV, and ROI

Businesses cannot afford systems that lose money. Economic feasibility answers:

"Will this system make money, and how long will it take?"

Key Metrics

1. Payback Period (PP): Formula: $$PP=\frac{\text{Initial Investment}}{\text{Annual Profit}}$$ Worked Example (Ncell’s New Billing System):

   • Cost: Rs. 2,00,000
   • Annual Profit: Rs. 50,000
   • PP = 2,00,000 / 50,000 = 4 years
   • But with 6% interest, use the discounted payback period: $$PP=\frac{2,00,000}{(50,000\times 0.943)}\approx 4.24\text{ years}$$

2. Net Present Value (NPV): Measures total profit over time, accounting for inflation. Formula: $$NPV=\sum _{t=0}^n\frac{{\text{Cash Flow}}_t}{(1+r{)}^t}-\text{Initial Cost}$$ Example (eSewa’s Online Payment System):

   • Initial Cost: Rs. 75,000
   • Annual Benefit: Rs. 85,000
   • Recurring Cost: Rs. 35,000/year
   • Discount Rate (r): 12%
   • NPV Calculation (for 5 years):

     Year 0: -75,000
     Year 1: (85,000 - 35,000)/1.12 ≈ 44,643
     Year 2: 50,000/1.12² ≈ 39,323
     ...
     Total NPV ≈ Rs. 1,20,000 (positive → **feasible**)
     

3. Return on Investment (ROI): $$ROI=\frac{\text{Net Profit}}{\text{Initial Investment}}\times 100\%$$ Example (Daraz’s Warehouse Automation):

   • Cost: Rs. 5,00,000
   • Annual Savings: Rs. 2,00,000
   • ROI = (2,00,000 / 5,00,000) × 100 = 40% per year

Decision Rule:

• NPV > 0 → Invest (e.g., Ncell’s 5G upgrade).
• PP < 3 years → Low risk (e.g., eSewa’s mobile app).
• ROI > 20% → Highly profitable (e.g., Daraz’s AI recommendations).

4. System Documentation: Guidelines for Forms and Reports

Poor documentation causes system failures (e.g., Kathmandu’s traffic lights malfunctioning due to unclear manuals). Follow these SMART rules:

Designing Forms (Input)

flowchart TD
    A["Start"] --> B["Keep it Simple"]
    B --> C["Meaningful Labels"]
    C --> D["Accurate Data Types"]
    D --> E["Relevant Fields Only"]
    E --> F["Timely Updates"]
    F --> G["End"]

Designing Reports (Output)

flowchart TD
    A["Start"] --> B["Clear Title"]
    B --> C["Logical Grouping"]
    C --> D["Consistent Formatting"]
    D --> E["Actionable Insights"]
    E --> F["End"]

Example (Bank Loan Report):

• Bad: Raw numbers like "Loan ID: 123, Amount: 5,00,000".
• Good:

  LOAN APPROVAL REPORT
  --------------------
  | Customer: Ram Prasad | Status: Approved |
  | Amount: Rs. 5,00,000 | EMI: Rs. 15,279/month |
  | Due Date: 2025-12-31 | Action: Sign documents |
  

5. System Implementation Methods

Choosing the wrong method can break a system (e.g., NTC’s direct cutover of the new billing system caused a 3-day outage). Compare:

Example (Ncell’s New Customer Portal):

• Method: Phased rollout
  • Phase 1: Login + profile (10% users).
  • Phase 2: Bill payment (50% users).
  • Phase 3: Full features (100% users).
• Why? Avoids overwhelming IT support with bugs.

6. CASE Tools in SDLC Phases

CASE tools replace manual work in each SDLC phase. Here’s how:

sequenceDiagram
    participant User
    participant UpperCASE as Upper CASE (e.g., ERwin)
    participant LowerCASE as Lower CASE (e.g., Visual Studio)
    participant System

    User->>UpperCASE: Draw ERD for Bank Database
    UpperCASE->>User: Generate SQL Schema
    User->>LowerCASE: Code in VS Code
    LowerCASE->>System: Deploy Loan System
    System-->>User: Run Selenium Tests

In the Real World

1. eSewa’s Maintenance Team

   • Perfective Maintenance: Reduced payment processing time from 8 seconds to 3 seconds using CASE tools (Lower CASE: Python optimizations).
   • Adaptive Maintenance: Updated KYC rules after Nepal Rastra Bank’s 2023 regulations (cost: Rs. 10M, saved Rs. 50M in fines).

2. Ncell’s 5G Rollout

   • Economic Feasibility:
     • Initial Cost: Rs. 10 billion.
     • NPV (5 years): Rs. 15 billion (ROI: 50%).
   • Implementation: Phased (started in Kathmandu, then Pokhara, then rural areas).
   • CASE Tools: Used IBM Rational for network design (Upper CASE).

3. Daraz’s AI Recommendations

   • Perfective Maintenance: Improved recommendation accuracy from 60% to 85% using machine learning models (coded in Python with Lower CASE tools).
   • Preventive Maintenance: Weekly Selenium tests to catch bugs before launch.

4. NTC’s Billing System Overhaul

   • Problem: Old system couldn’t handle GST changes.
   • Solution:
     • Upper CASE (ERwin): Redesigned data model.
     • Implementation: Parallel run for 3 months to compare old vs. new bills.
   • Cost: Rs. 30M, saved Rs. 200M in manual processing.

Exam Tip

1. Payback Period Questions:

   • Always discount cash flows if interest is given.
   • Example: For Rs. 2,00,000 investment with Rs. 50,000/year profit and 6% interest: $$PP=\frac{2,00,000}{50,000\times 0.943}\approx 4.24\text{ years}$$

2. CASE Tools:

   • Upper CASE = Design phase (ERD, DFD).
   • Lower CASE = Coding/testing phase (Visual Studio, Selenium).
   • I-CASE = Full SDLC (IBM Rational).

3. Documentation:

   • Forms: Follow SMART rules (Simple, Meaningful, etc.).
   • Reports: Must have clear titles, logical grouping, and actionable insights.

4. Implementation Methods:

   • Direct cutover = High risk, low cost.
   • Parallel = Safe but expensive.
   • Phased/Pilot = Best for large systems (e.g., NTC, banks).

5. Maintenance Types:

   • Corrective = Bug fixes (e.g., eSewa crashes).
   • Adaptive = Legal changes (e.g., Ncell’s 5G rules).
   • Perfective = Performance upgrades (e.g., Daraz’s speed).
   • Preventive = Security patches (e.g., banks’ fraud tests).

Common Mistakes to Avoid:

• Forgetting to discount cash flows in PP/NPV questions.
• Drawing DFDs without balancing I/O (exam favorite!).
• Confusing Upper CASE (design) with Lower CASE (coding).
• Ignoring recurring costs in NPV calculations.