Server Hardware Strategies & Management: Types, Procurement, Risk & Maintenance

TAKEAWAYS:

• Server types (web, file, mail, database, proxy, etc.) are specialized for specific roles in enterprise networks, each requiring distinct hardware configurations.
• Hardware strategies include redundancy (RAID, hot-swappable components), scalability (blade servers, modular designs), and energy efficiency (PUE, virtualization).
• Procurement planning follows a structured approach: needs assessment → vendor evaluation → budgeting → risk mitigation (e.g., SLAs, warranties).
• Risk management in server hardware involves physical threats (fire, theft), hardware failures (disk crashes, overheating), and supply chain risks (vendor reliability).
• Maintenance includes preventive (regular cleaning, firmware updates) and corrective (RMA, replacement) actions, with documentation as a critical component.
• Cost-benefit analysis balances upfront hardware costs against long-term TCO (Total Cost of Ownership), including energy, cooling, and downtime risks.

---

1. Types of Servers and Their Hardware Requirements

Servers are specialized computers designed to handle specific tasks in a network. Their hardware must align with their role to ensure performance, reliability, and security.

Common Server Types and Hardware Needs

Worked Example: NEPSE’s Trading Server Hardware

Nepal Stock Exchange (NEPSE) requires ultra-low latency and high availability for its trading platform. Their server hardware includes:

• Dual Intel Xeon Platinum CPUs (for real-time order matching).
• 1TB+ RAM (to handle thousands of concurrent trades).
• RAID 10 storage (for transaction logs and audit trails).
• Redundant power supplies (2x PSUs) and hot-swappable fans.
• 10Gbps NICs (to connect to exchange participants).

Why?

• Latency: Multi-core CPUs and NVMe SSDs reduce delay in order execution.
• Reliability: RAID 10 ensures no data loss if a disk fails.
• Scalability: Additional NICs allow future expansion to 40Gbps.

---

2. Hardware Strategies for Servers

Hardware strategies ensure servers meet performance, reliability, and cost-efficiency goals.

A. Redundancy and Fault Tolerance

Redundancy prevents single points of failure. Key techniques:

• RAID Levels (for storage redundancy):

• Hot-Swappable Components:

  • Power Supplies (PSUs): Replaceable without shutdown.
  • Fans: Prevent overheating (critical in data centers).
  • Hard Drives: RAID arrays allow drive replacement while running.

• Uninterruptible Power Supply (UPS):

  • Provides backup power during outages (e.g., Ncell’s data centers use UPS to avoid service disruption during load shedding).

B. Scalability Strategies

Servers must grow with demand. Common approaches:

1. Vertical Scaling (Scaling Up):
   • Adding more CPU cores, RAM, or faster storage to a single server.
   • Example: Upgrading a Daraz order-processing server from 32GB RAM to 128GB during Diwali sales.
2. Horizontal Scaling (Scaling Out):
   • Adding more servers (e.g., load balancers distributing traffic across multiple web servers).
   • Example: Khalti’s payment gateway uses multiple servers to handle peak transactions during Dashain.

3. Modular Servers (Blade Servers):
   • Definition: Servers with removable blades (e.g., HP BladeSystem).
   • Advantages:
     • Shared power/cooling reduces costs.
     • Easy to add/remove blades.
   • Disadvantage: Single point of failure if the chassis fails.

C. Energy Efficiency and Cooling

Data centers consume ~1-2% of global electricity. Strategies to reduce costs:

• Power Usage Effectiveness (PUE):
  • Formula: PUE = Total Facility Power / IT Equipment Power
  • Goal: PUE < 1.2 (best-in-class).
• Cooling Methods:
  • Air Cooling: Fans + cold aisles (used in most small data centers).
  • Liquid Cooling: Immersion cooling (used by Google’s data centers).
  • Free Cooling: Using outside air (e.g., NTC’s fiber optic nodes in cooler climates).

---

3. Server Procurement Planning

Procurement is a structured process to select the right hardware while managing risks.

Step-by-Step Procurement Plan

A. Needs Assessment

• Questions to Answer:
  • What is the server’s role? (e.g., eSewa’s authentication server needs high security).
  • What is the expected workload? (e.g., Pathao’s ride-matching server requires low latency).
  • What are compliance requirements? (e.g., PCI-DSS for banks).

B. Vendor Evaluation

C. Budgeting

• Cost Components:
  • Upfront Costs: Server hardware, racks, UPS.
  • Operational Costs: Electricity, cooling, maintenance.
  • Hidden Costs: Downtime, data loss, upgrades.
• Example Budget for a Small Enterprise:

  Item	Cost (USD)
  Dell PowerEdge Server	$3,500
  1TB NVMe Storage	$800
  Rack Mounting	$200
  UPS (1kVA)	$500
  Total	$4,000

D. Risk Management

Worked Example: Kathmandu Traffic Management System

• Risk: Server failure during peak hours (e.g., Dashain) causes traffic gridlock.
• Solution:
  • Redundant servers (active-passive setup).
  • Battery-backed UPS for 30-minute runtime.
  • Cloud backup of traffic data (AWS/Nepal Cloud).

---

4. Server Hardware Maintenance

Maintenance ensures longevity, performance, and security.

A. Preventive Maintenance

B. Corrective Maintenance

• RMA (Return Merchandise Authorization):
  • Process for replacing faulty hardware (e.g., dead motherboard in a Pathao server).
• Hardware Replacement:
  • Example: Replacing a failed 10Gbps NIC in a Ncell core router.
• Documentation:
  • Log all maintenance in a CMDB (Configuration Management Database).

---

In the Real World

1. Ncell’s 4G/5G Core Network:

   • Hardware Strategy: Uses Cisco ASR 9000 routers with hot-swappable line cards for redundancy.
   • Why? Ensures 99.999% uptime (5 nines) for voice/data services.
   • Real Example: During the 2023 monsoon floods, Ncell’s redundant power systems kept the network running despite power outages in Kathmandu.

2. eSewa’s Payment Gateway Servers:

   • Hardware: Dell PowerEdge R750 with RAID 10 storage and TLS 1.3 acceleration cards.
   • Why?
     • RAID 10 prevents data loss if a disk fails during high transaction volumes (e.g., Dashain).
     • TLS acceleration speeds up encrypted payments (critical for PCI-DSS compliance).

3. Nepal Stock Exchange (NEPSE) Trading Servers:

   • Hardware: Custom-built servers with FPGA accelerators for ultra-low-latency trading.
   • Why?
     • FPGAs reduce order execution time from 10ms to <1ms.
     • Redundant NICs ensure no network bottleneck during high-frequency trading.

---

Exam Tip

This unit is heavily tested on:

1. Server Types & Hardware Matching (5-mark questions):

   • Example: "Which server type requires the highest RAM and why?" → Database server (for caching query results).
   • Key Point: Memorize the hardware requirements for each server type (CPU, RAM, storage, NIC).

2. Procurement Planning (4+4-mark questions):

   • Structure Your Answer:
     1. Needs Assessment (1 mark) → Define server role and workload.
     2. Vendor Selection (2 marks) → Compare Dell vs. HP vs. local vendors.
     3. Budgeting (1 mark) → Breakdown of costs (hardware + operational).
     4. Risk Management (2 marks) → Redundancy, SLAs, backup plans.
   • Example Question: "Prepare a procurement plan for a file server handling 1TB of student records for TU."
     • Answer Tip: Mention RAID 6 for storage, UPS for power, and local vendor (Nepal Data Center) for faster support.

3. Hardware Strategies (5-mark questions):

   • Common Pitfalls:
     • ❌ Saying "RAID 0 is best for reliability" → Wrong! RAID 0 has no redundancy.
     • ❌ Ignoring cooling in data center design → Always mention PUE or hot/cold aisles.
   • Example Question: "What are the essential hardware strategies for a web server hosting a Nepali news website?"
     • Answer:
       • Redundancy: RAID 1 for OS, RAID 10 for user uploads.
       • Scalability: Load balancer + multiple web servers.
       • Cooling: Air conditioning with PUE < 1.5.
       • Backup: Daily cloud backups (AWS S3).

---

Final Checklist for Full Marks: ✅ Define key terms (RAID, PUE, hot-swappable). ✅ Compare at least two hardware strategies (e.g., RAID 5 vs. RAID 10). ✅ Apply to a real-world scenario (e.g., Ncell, eSewa, NEPSE). ✅ Use visuals (tables, diagrams, mermaid flows) to explain complex ideas. ✅ Structure answers clearly (e.g., procurement plan steps).