📡Systems Approach to Computer Networks Unit 3 – Network Core: Switching and Structure

Key Concepts

• Switching enables efficient communication between devices on a network by directing data packets to their intended destinations
• Switches operate at the data link layer (Layer 2) of the OSI model and use MAC addresses to forward frames
• Network structure refers to the physical and logical arrangement of devices, while topology describes the geometric layout of the network
• Latency, throughput, and bandwidth are critical performance metrics for evaluating network efficiency and responsiveness
• Quality of Service (QoS) mechanisms prioritize traffic to ensure reliable delivery of time-sensitive applications (voice, video)
• Redundancy and failover mechanisms ensure high availability and minimize downtime in case of device or link failures
• Network management tools (SNMP, NetFlow) enable monitoring, configuration, and troubleshooting of switches and other network devices

Network Switching Basics

• Switching involves receiving data packets on one port, examining the destination MAC address, and forwarding the packet to the appropriate output port
• MAC address table (CAM table) maintains a mapping of MAC addresses to switch ports, enabling efficient packet forwarding
• Switches learn MAC addresses dynamically by examining the source MAC address of incoming packets and associating them with the corresponding port
• Broadcast domains are created by switches, limiting the scope of broadcast traffic to devices connected to the same switch or VLAN
• VLANs (Virtual LANs) logically segment a switch into multiple broadcast domains, improving security and performance by isolating traffic
• Switches use STP (Spanning Tree Protocol) to prevent loops in the network topology by blocking redundant paths
• Port mirroring (SPAN) allows traffic from one or more ports to be copied and sent to a monitoring port for analysis or troubleshooting

Types of Switches

• Layer 2 switches operate at the data link layer and forward packets based on MAC addresses
• Layer 3 switches (multilayer switches) perform both Layer 2 switching and Layer 3 routing, enabling inter-VLAN communication and advanced features (ACLs, QoS)
• Managed switches offer advanced configuration options and monitoring capabilities through a web interface, CLI, or SNMP
  • Enable granular control over port settings, VLANs, QoS, and security features
  • Support link aggregation (LAG) for increased bandwidth and redundancy
• Unmanaged switches are plug-and-play devices with fixed configurations and limited features, suitable for small networks or edge deployments
• PoE (Power over Ethernet) switches provide both data connectivity and electrical power to devices (IP phones, wireless APs) through Ethernet cables
• Stackable switches can be interconnected to form a single logical switch, simplifying management and increasing port density
• Modular switches offer flexibility through interchangeable line cards and support for high-density deployments

Switch Architecture

• Shared memory architecture uses a central memory buffer to store packets, with the switching fabric connecting input and output ports
  • Offers low latency and good performance for small to medium-sized switches
  • May experience contention and reduced throughput under heavy loads
• Crossbar switch architecture provides a dedicated path between each input and output port, enabling parallel packet processing
  • Delivers high throughput and scalability for larger switches and high-performance applications
  • Requires more complex scheduling algorithms to avoid internal blocking
• Distributed switch architecture employs multiple switching fabrics and line cards, each with its own processing capabilities
  • Enables modular scalability and improved fault tolerance
  • Requires careful design to ensure consistent performance and avoid bottlenecks
• ASICs (Application-Specific Integrated Circuits) are specialized hardware components optimized for high-speed packet processing and forwarding
• Network processors offer programmability and flexibility for implementing advanced features and protocols
• Buffering techniques (input buffering, output buffering, shared memory) help manage congestion and optimize switch performance

Switching Protocols

• Ethernet is the most widely used Layer 2 protocol for wired networks, defining frame formats, MAC addressing, and media access control
• Spanning Tree Protocol (STP) prevents loops in switched networks by blocking redundant paths and ensuring a single active path between any two devices
  • RSTP (Rapid STP) provides faster convergence and improved recovery times compared to traditional STP
  • MSTP (Multiple STP) allows multiple spanning tree instances for better utilization of redundant paths
• Link Aggregation Control Protocol (LACP) enables the bundling of multiple physical links into a single logical link for increased bandwidth and redundancy
• VLAN Trunking Protocol (VTP) simplifies VLAN configuration management by propagating VLAN information across multiple switches
• IGMP (Internet Group Management Protocol) manages multicast group membership and enables efficient distribution of multicast traffic
• Jumbo frames are Ethernet frames with a payload larger than the standard 1500 bytes, reducing overhead and improving throughput for bulk data transfers
• Port security protocols (802.1X, MAC address filtering) authenticate and authorize devices connecting to switch ports, enhancing network security

Network Structure and Topology

• Hierarchical network design organizes the network into distinct layers (access, distribution, core), improving scalability, performance, and manageability
  • Access layer connects end devices and provides local switching and access control
  • Distribution layer aggregates traffic from access layer switches and performs inter-VLAN routing and advanced services
  • Core layer provides high-speed backbone connectivity and interconnects distribution layer switches
• Star topology connects all devices to a central switch, offering simplicity, centralized management, and easy expansion
  • Single point of failure at the central switch can impact the entire network
• Bus topology connects devices along a single cable, with data transmitted in both directions
  • Easy to install and extend, but prone to collisions and single points of failure
• Ring topology connects devices in a closed loop, with data transmitted unidirectionally
  • Offers redundancy and fault tolerance, but requires specialized protocols (Token Ring) to manage access and prevent loops
• Mesh topology establishes multiple connections between devices, providing high redundancy and fault tolerance
  • Full mesh topology connects every device to every other device, offering maximum resilience but high complexity and cost
  • Partial mesh topology selectively interconnects devices based on traffic patterns and criticality, balancing redundancy and efficiency

Performance and Optimization

• Bandwidth refers to the maximum amount of data that can be transmitted over a link or network segment per unit of time (bps, Mbps, Gbps)
• Latency is the time taken for a packet to travel from source to destination, including processing, queuing, and propagation delays
  • Minimizing latency is crucial for time-sensitive applications (VoIP, video conferencing, high-frequency trading)
• Throughput measures the actual amount of data transferred over a link or network segment per unit of time, considering overhead and congestion
• Quality of Service (QoS) mechanisms prioritize and manage traffic based on application requirements and network conditions
  • Classification and marking assign priority levels to packets based on criteria (IP address, port number, DSCP)
  • Queuing and scheduling algorithms (WRR, PQ, CQ) determine the order in which packets are processed and transmitted
  • Congestion avoidance techniques (WRED, ECN) proactively manage buffer utilization and prevent network congestion
• Link aggregation (LAG, EtherChannel) combines multiple physical links into a single logical link, increasing bandwidth and providing link-level redundancy
• Load balancing distributes traffic across multiple paths or devices to optimize resource utilization and improve performance
  • Equal-cost multi-path (ECMP) routing forwards packets across multiple equal-cost paths to the same destination
  • Port-channel load balancing distributes traffic across the physical links within a LAG based on criteria (MAC address, IP address, TCP/UDP port)
• Jumbo frames reduce overhead and improve throughput by increasing the payload size, but require end-to-end support and careful network design

Real-World Applications

• Data center networks employ high-performance switches (10G, 40G, 100G) and architectures (leaf-spine, Clos) to support virtualization, cloud computing, and big data workloads
  • Top-of-Rack (ToR) switches provide connectivity to servers and storage within a rack
  • End-of-Row (EoR) switches aggregate traffic from multiple racks and connect to the core layer
• Campus networks use a hierarchical design with access, distribution, and core layers to connect buildings, departments, and user devices
  • Access layer switches provide connectivity and PoE for end devices (PCs, IP phones, APs)
  • Distribution layer switches perform inter-VLAN routing, QoS, and security policy enforcement
  • Core layer switches interconnect distribution switches and provide high-speed backbone connectivity
• Industrial Ethernet networks adapt switching technologies for harsh environments and real-time control systems
  • Rugged switches with extended temperature ranges and vibration resistance
  • Support for deterministic protocols (EtherCAT, PROFINET) and time-sensitive networking (TSN) standards
• Software-Defined Networking (SDN) decouples the control plane from the data plane, enabling centralized network management and programmability
  • OpenFlow protocol enables communication between SDN controllers and switches
  • Network Function Virtualization (NFV) replaces dedicated hardware appliances with virtualized network functions running on commodity servers
• Wireless LAN (WLAN) controllers manage and configure wireless access points (APs) in enterprise networks
  • Centralized management, security, and QoS policies for wireless clients
  • Integration with wired network infrastructure through switched ports and VLANs