Comprehensive Study Notes on Network Architecture and Requirements Analysis
Introduction to Network Analysis, Architecture, and Design
The primary objective of these processes is to enable intelligent, informed network engineering decisions. This applies to any ICT Engineering project, ranging from developing national networks to small enterprise Local Area Networks (LANs) and overall network upgrades.
Familiarity with basic networking concepts is required, including the TCP/IP Protocol Suite and the basics of network routing, performance, and management.
The methodology provides step-by-step procedures to:
Gather, derive, define, and validate real requirements.
Evaluate and select vendors, products, and services.
Develop traceability between requirements, architecture decisions, and design decisions.
Determine where to apply routing protocols and IP addressing mechanisms.
Determine where to apply performance mechanisms, including Quality of Service (QoS), Service Level Agreements (SLAs), and policies.
The Network Lifecycle Processes
Network Analysis: The process of learning what users, applications, and devices need from the network. It involves listening to users and their needs to understand the system and the problems being solved. It leads to the documentation of an audit trail (data, decisions, etc.).
Network Architecture: This process uses analysis information to develop a conceptual, high-level, end-to-end network structure. It involves making technology and topology choices and determining relationships among network functions such as addressing, routing, security, network management, and performance. There is rarely a single "right" architecture; multiple solutions may work.
Network Design: This provides the physical details of the architecture, including blueprints and drawings. It involves selecting vendors, service providers, equipment types, and configurations. Design involves optimizing the system across relationships and making trade-offs (e.g., cost vs. performance, simplicity vs. function).
Network Hierarchy and Diversity
Hierarchy: Provides separation of the network into segments, such as separate networks, subnets, and broadcast domains. Hierarchy becomes necessary as traffic grows beyond capacity or results in congestion.
Diversity (Redundancy/Interconnectivity): While hierarchy defines structure, diversity balances it by interconnecting the network at different levels to enhance performance.
Dynamics: The interaction between hierarchy and diversity is a fundamental trade-off in network architecture and design.
Systems Methodology and Service Description
Systems Methodology: This approach views the network and a subset of its environment as a single system. It emphasizes the services each network will provide and support to determine the network's characteristics and capabilities.
A System: Defined as a set of components working together to support or provide connectivity, communications, and services to users. Identifying components helps in understanding how they interface across boundaries.
Network Services: Levels of performance and function. They can be viewed as services offered by the network to the system (devices, applications, users) or as requirements expected from the network.
Service Levels: Grouped service characteristics (performance and functional parameters) used to make provisioning, billing, and accounting easier. They are distinguished by degrees of predictability and performance.
Example: A premium service level might combine a capacity of and reliability of uptime.
Performance Characteristics and Metrics
Service Performance Characteristics:
Capacity: Includes bandwidth, throughput, and goodput.
Delay: Includes end-to-end delay, round-trip time, latency, and jitter.
RMA (Reliability, Maintainability, and Availability):
Reliability: A statistical indicator of the frequency of failure.
Maintainability: A statistical measure of the time to restore the system to fully operational status after a fault.
Availability: The relationship between mission-critical failures and the time to restore service.
Service Predictability Categorization:
Best-Effort: Unpredictable and unreliable delivery with no control over how the network satisfies the request.
Predictable: Provides some degree of predictability between best-effort and guaranteed.
Guaranteed: The opposite of best-effort; provides high-performance, predictable outcomes.
Performance Envelopes: Combinations of two or more performance requirements with threshold limits. They help visualize the regions (e.g., low and high performance) in which the network is expected to operate.
2D Envelope: Typically plots capacity (data sizes) against end-to-end delay.
3D Envelope: Plots capacity, delay, and RMA.
Requirements Analysis: Core Concepts
Definitions: Requirements are descriptions of network functions and performance needed for the network to successfully support its users, applications, and devices.
Categorization and Prioritization:
Core/Fundamental Requirements: Essential for the success of the project.
Metrics: One or more measurements associated with each core requirement.
Features: Non-essential requirements.
Classification Flow: Requirements are gathered and derived from users, management, and staff. Through analysis, they are classified into: Core Requirements, Features, Requirements for Future Revision/Upgrades, Rejected Requirements, or Informational Requirements.
Relative Importance Key Words (RFC 2119):
Must / Shall / Required
Must Not / Shall Not
Should / Recommend
Should Not / Not Recommended
May / Optional
User, Application, and Device Requirements
User Requirements: The highest, least technical, and most subjective layer. These are qualitative requirements gathered from users to accomplish tasks. Key types include: timeliness, interactivity, reliability, presentation quality, adaptability, security, affordability, functionality, supportability, and future growth.
Application Requirements: Requirements needed for applications to operate successfully. User requirements have a direct impact here.
Application Types:
Mission-critical: Applications requiring predictable, guaranteed, and/or high-performance RMA.
Rate-critical: Applications requiring predictable, guaranteed, and/or high-performance capacity.
Real-time and Interactive: Applications requiring predictable, guaranteed, and/or high-performance delay.
Delay Classifications:
Real-Time: Strict timing relationship between source and destination.
Interactive: Loose timing relationship (Interactive-Burst, Interactive-Bulk).
Non-Real-Time: Asynchronous timing.
Application Groups: Telemetry/Command-and-Control, Visualization, Distributed Computing, Web Development, Bulk data transport, TeleService, Operations/Administration/Maintenance (OAM&P), and Client-Server.
Device Requirements: The network must support specific device types and performance characteristics.
Device Categories: Generic computing devices, Servers, Specialized devices.
Performance Characteristics: Storage performance (flash, disk-drive, tape), Processor, Memory (access times), Bus (capacity and arbitration), OS (protocol stack and APIs), and Device drivers.
Network Requirements, Management, and Security
Existing Networks: Designs must accommodate dependencies from the current infrastructure, including scaling, location dependencies, performance constraints, and interoperability.
Network Management: Four primary task categories:
Monitoring for event notification.
Monitoring for metrics and planning.
Network configuration.
Troubleshooting.
Management Performance: Considerations include monitoring methods (In-band vs. out-of-band), instrumentation methods (protocols), and centralized vs. distributed monitoring.
Security Risk Assessment: Conducted using a Risk Assessment Matrix. Risks are assessed based on Effect (A: Destructive, B: Disabling, C: Disruptive, D: No Impact) and Probability (A: Certain, B: Unlikely, C: Likely, D: Impossible).
Areas assessed: Users, Devices, Servers, Network Elements, Software Services, Data.
Threats: Unauthorized Access, Unauthorized Disclosure, Denial of Service, Theft, Corruption, Viruses, Physical Damage.
Operational Suitability and Supportability
Supplemental Requirements: Financial requirements (affordability) and Enterprise requirements (consolidating phone, fax, voice, video).
The Three Elements of Post-Implementation Supportability:
Operations: Ensuring proper management and maintenance identification.
Maintenance: Preventive and corrective actions, including parts, tools, and plans.
Human Knowledge: Documentation, training, and skilled personnel.
Deliverables: Specification and Map
Requirements Specification: A document listing and prioritizing gathered requirements. It includes columns for ID/Name, Date, Type, Description, Gathered/Derived source, Locations, Status, and Priority.
Example entry: Database application (DB1) requires a minimum of per session.
Requirements Map: Shows location dependencies between applications and devices across floors or buildings. It notes specific needs like "Visualization Application with Delay" or "GigE NICs" and user distributions (e.g., 60 engineers, 30 manufacturing users).
LAN distance examples: between campuses.