ITEC 3210 Notes York University Lecture 1 Network basics Classification Structured design Component view Layer view (Based on Network models) Indicators of success in network design (or technical goals) Organizational networks: importance & related jobs Network analysis & design: content, process & challenges Applications exercises on tech goals Partially covered: Standards, Future Trends Data Communications Data Communications: movement of computer info from 1 point 2 another by means of electrical or optical transmission systems Telecommunications: includes the transmission of voice & video to transmit data, usually for longer distances Networks: computers & other devices connected by some type of transmission media Scale: LAN – BN – WAN Access: internet, intranet & extranet Network Types Local Area Networks (LAN): covers a small and defined area; maybe one floor or one room o When LAN uise wireless circuits, they are called WLAN Backbone Networks (BN): High speed networks connecting other networks together; may span 100s of meters 2 several kms Wide Area Networks (WAN): Largest in scope, composed of leased circuits, may span 100s of km Intranet: a network (often a LAN) that uses Internet to share info within the org; only open for those inside the org; ex: employees accessing budgets & payroll info Extranet: shares info between orgs, accessing info of one company inside another company’s database, ex: checking another company’s inventory’ Structured Design aka BN architecture Block 1 = Access Layer: connects computers & printers to form a LAN Block 2 = Distribution Layer: connects LANs to form BNs (or access other layers) Block 3 = Core layer i.e., BN that interconnects different BNs o ONLY used in networks across 2 or more adjacent buildings Ex: 1 ITEC 3210 Notes York University Network Architecture LAN Distribution layer or building Core layer or campus BN Data centre Enterprise edge o WAN/MAN o Internet access o E-commerce edge Networks may have an architecture that requires only some of these components Basic Network Components Client: user to access network & receive data from server; ex: pc, laptops, cell phone Server: stores & transmits data 2 client; ex: web server, mail (gmail) Circuit: pathway/connection between client & server; ex: fiber optic cable Components of Networks: Hardware Computers: Mainframes, servers, workstations, personal computers, terminals Connectivity devices (AKA intermediary devices) o Access layer: switches & possible access points o Distribution & core layer: switches but also routers & other Transmission media or circuits: guided (cables + connectors) or wireless Other: NICs, repeaters, printers, storage devices Components of Networks: Software Application Software Network Operating Systems Protocols & standards o Agreed upon rules 4 transmitting data between 2 (or more) devices o Some of these rules r implemented in hw Multi-layer view of networks Models that reduce the complexity of data communication between computers Network models adopt a modular approach: o Process of transferring data is divided into layers o Each layer handles specific sub-tasks w specific protocopls Comm occurs: o Between adjacent layers on the same lvl (vertical) 2 ITEC 3210 o Notes York University Between corresponding layers on diff systems (horizontal) Pros & cons of using layers Advantages o Easier sw developers + standards that define the comm rules o Networking functions r modular (can b changed at anytime) o Easier 2 troubleshoot or make changes 2 one layer at a time o App devs only need 2 worry about the app layer in their programs Disadvantages o Inefficient because the encapsulation/de-encapsulation at each layer needs processing o Encapsulation into a protocol data unit (PDU) increases overhead at each layer Network Models Lecture 2 – Application layer Key Technical Goals in Network Design 1. Performance: Measured by response time or throughput. 2. Availability: Ensuring the system is operational and accessible. 3. Scalability: The ability to handle an increasing number of users or data without significant performance degradation. These goals guide the selection of application architectures and best practices. Context of the Application Layer The Application Layer defines how applications work in a network, including protocols, data formats, and other requirements. It is primarily the responsibility of software developers, not network designers. 3 ITEC 3210 Notes York University Network designers, however, decide which application architecture to use for a specific application, based on the five functions of application programs: 1. Data storage: Storing and retrieving data. 2. Data access logic: Querying and accessing subsets of data. 3. Application logic: The core work performed by the application. 4. Presentation logic: The user interface. 5. Services logic: Automated access to other applications (if needed). The choice of architecture is justified using the three technical goals mentioned above. Types of Network Applications Network applications are categorized into two main types: 1. Transaction Processing Systems (TPS): o Focused on capturing and processing business transactions (e.g., receiving orders, payments, etc.). o Examples: Airline booking systems, order management systems, accounting systems, and university enrollment systems. o TPS applications require a database server and a Database Management System (DBMS). o TPS is the primary focus of this lecture. 2. Non-TPS Applications: o Examples: Email, web browsing, file storage, file transfer, and videoconferencing. o These applications do not require a database server or DBMS. Application Architectures The application architecture defines how the five functions of the application layer are distributed among computers (clients and servers). It is distinct from network architecture. The lecture covers five main architectures: 1. Host-Based Architecture The host/server (usually a mainframe) performs all five functions. The client (a terminal or microcomputer) sends and receives messages to/from the host. Limitations: o Expensive and difficult to upgrade incrementally. o Single bottleneck: Overload on the server can increase response time. Typical Use: Large organizations (e.g., banks, insurance companies, government agencies) that process large volumes of data. 2. Client-Based Architecture 4 ITEC 3210 Notes York University Became popular in the late 1980s with the rise of PCs and LANs. The server handles only data storage, while the client handles data access, application, and presentation logic. Limitations: o Heavy network load: All data must travel to the client for processing. o Limited processing power and scalability on the server side. o Poor network performance. Best Practices: Generally considered a bad practice due to inefficiencies. 3. Client-Server (C-S) Architecture The most common architecture today. The server handles data storage, data access logic, and (optionally) application logic. The client handles presentation logic and (optionally) application logic. Advantages: o Distributed processing improves network performance. o Allows hardware/software from different vendors to work together. o Reduces unnecessary traffic by keeping data access logic on the server. Disadvantages: o Configuring hardware/software from different vendors can be challenging. Variations of C-S Architecture: Two-Tier Architecture: o Application logic is split between one client and one server. o Common for small-scale systems with fewer than 50 users. Three-Tier Architecture: o The client handles presentation logic. o An application server handles application and services logic. o A database server handles data storage and access logic. n-Tier Architecture: o More than three types of computers are used (e.g., adding a web server). o Advantages: Load balancing and scalability. o Disadvantages: Increased network load and complexity. Thin vs. Thick Clients: Thin Clients: o Minimal application logic resides on the client. o Easier to manage and cheaper (no storage device required). Thick Clients: o Most or all application logic resides on the client. o Necessary for users who need to install their own applications. 4. Cloud Computing Architecture 5 ITEC 3210 Notes York University A variation of the C-S architecture where infrastructure (hardware and software) is outsourced to a third party. Cloud Computing Models: 1. Software as a Service (SaaS): Third party provides all hardware and software. Example: Gmail. 2. Platform as a Service (PaaS): Third party provides programming tools, OS, and hardware. Users develop and own their applications. Example: Microsoft Azure. 3. Infrastructure as a Service (IaaS): Third party provides hardware; users provide software. Example: Amazon Web Services (AWS). 5. Peer-to-Peer (P2P) Architecture All computers act as both clients and servers. Common in file-sharing applications but not widely used in organizations. Example: P2P file-sharing systems. Best Practices for Selecting an Architecture Host-Based Architecture: Suitable for large organizations processing high volumes of data (e.g., TPS). Client-Server Architecture: Most applications today use this architecture, with thin or thick clients depending on the use case. Cloud Computing: Used when outsourcing infrastructure, but risks related to data security and privacy must be considered. The selection of an architecture should be based on the three technical goals (performance, availability, scalability) rather than cost alone. Middleware in Client-Server Architecture Middleware acts as an intermediary between client and server applications. Functions: o Translates between software from different vendors. o Manages message transfers. o Insulates network changes from clients (e.g., adding a new server) 6 ITEC 3210 Notes York University Lecture 3 – Physical Layer & data link layer Content: Physical layer o Ciruits o Comm media and relevant considerations o Key aspects of data transmission Data link layer o Media access control o Error control o Ethernet protocol Context & Applicability Physical layer: bit transmission over circuits in LANs, WLANs, BNs, MANs & WANs Data link layer focuses on Media Access Controls (MACs), message delineation, message delivery (device-2-device) & error control 7 ITEC 3210 Notes York University The physical layer Circuits o Physical: canbles or wl media connecting network components o Logical: transmission characteristics of the circuit Logical circuits possible configs: o Point to point o Multipoint Data flow: circuits can b designed 2 be uni or bi-directional o Simplex: data flows one way o Half duplex: flows both directions but one at a time o Full duplex: data flows simultaneously in both directions Circuits Point 2 point circuits include most wired connections today Multipoints are most commonly used in wireless Shared circuits r usually less expensive Media Definition: physical matter used 2 carry voice or data transmissions Guided media: transmission goes along physical medium Wireless (radiated) media: transmission flows thru the air Guided media 8 ITEC 3210 Notes York University Twisted-pair (TP) cable: insulated pairs of wires bundled together; wires r twisted 2 reduce electromagnetic interference o Sometimes use additional shielding (STP) o Grades include cat 5e, cat6, cat 6a etc… to support higher nominal data rates NDR) o Characteristics Price: inexpensive Distance – typically up to 100m (without repeaters) – due 2 attenuation; limit is part of our standards Uses: telephones, LANs Coax cable: single copper core, + outer insulation, shielding & inner insulation; less prone 2 interference o Characteristics Cheap but more costly than TP Distance: up to 2km Use: cable tv, internet Thick, harder 2 install Fiber optic: optical core made of glass or plastic; data is transmitted using light from lasers of LEDs; resistance 2 interference & corrosion; capable of extremely fast data rates o Characteristics Expensive 500m-100+km Used 4 trunk line/backbone, long distance circuits More secure than TP o Types/grades Multimode (50 micron core – 62.5 typ, ;500m max) Graded index multimode (1km max) Single mode (about 5-7 micron core; 100 km or more) Wireless Media Microwave: high freq radio comm; needs line of sight that might need big antennas or towers; strongly affected by the weather o About 60km; uses trunk line/ backbone, long distance Satellite: special form of microwave comm o Long distances may lead 2 propagation delays (GSO bad; LEO OK) o NOT commonly used in org networks Communication media: standards 9 ITEC 3210 Notes York University Structured cabling standard (p 64) o In line w structured design o Possible app Factors 2 consider in media selection Type of network o LANs & BNs: TP cable o WLANs: radio waves Cost o TP is cheaper for networks o WLANs’s radio waves have reasonable price Transmission distance: attenuation (+ dispension (for fiber)) o With no repeater, 100m for TP, 500m to 100km for fiber o Radio waves: around 100m to 150m Security o Wireless is less safe since msgs can b intercepted o Fiber is safer than TP Error rates: highest for wl media, lowest for fiber Data rates supported: Typically guided media support higher data rates than radio waves Digital Transmission of Digital Data Computers produce digital (binary) data A coding scheme (aka code) like ASCII or Unicode helps represent char using bits (8=16 bits p character) Transmission modes o How bits of a single character travel o Parallel (one-several bytes at a time) vs serial (one bit at a time) transmission Digital transmission o Sending and receiving components have to agree on rules to follow including voltage levels 2 represent “1” and “0” o Those rules are defined by the standard that the devices are compatible with; network designers need to ensure compatibility between all devices in the network o Ethernet is the most used network protocol that controls how data is transmitted over a wired LAN & is referred 2 as the IEEE 802.3 protocol 10 ITEC 3210 Notes York University How Ethernet transmits data o Transmission: serial for most earlier 10 Mb & 100Mb TP ethernet, parallel 4 higher NDR o Signaling technique (i.e., method of transmitting bits) depends on the version used o Key implication for network design: multi-speed devices may be required Analog Transmission of Digital Data Telephone system was originally built for analog data (the message is unstable and the voice is like a mimic) Analog transmission: signal is sent on telephone networks or other analog networks A modem converts the digital data produced by computers into an analog signal for transmission over analog voice-grade circuits at the receiving end, a second modem converts the analog signal into digital signals again Digital Transmission of Analog Data Somewhat “opposite” to that from analog; A codec device or software at the source converts an analog voice signal into digital signal for transmission over a digital circuit and a codec at the receiving end translates the digital signal back into analog data Application example: VoIP o VoIP phones have built-in codecs o Phone is connected to LAN (i.e., no specific wiring needed) o Requirements re. other equipment (switch, voice gateway, etc.) depend on implementation (see right) The Data Link Layer Media Access Control (1/2) Refers to rules to decide which computer transmits and when i.e., decide if 2 computers on the same access layer can send data at the same time 2 possible situations in today’s computer networks: 11 ITEC 3210 Notes York University o In the case of multi-point (i.e., shared) circuits, simultaneous transmission is NOT allowed, so control is needed (case of access points and hubs) o In the case of full-duplex point-to-point circuits, simultaneous transmission is allowed, so no control is needed (case of switches and routers) o Question: Which of the 2 situations is a “best practice”? Why? (more on this in Ch 7) Two approaches to control in current technologies: o Contention access: Transmit whenever circuit is available with no centralized control o Controlled access: Transmit whenever circuit is available with centralized control Contention o Transmit whenever circuit is available with no centralized control o Common in Ethernet LANs o When devices transmit at the same time, a collision occurs o E.g., CSMA/CD the Ethernet control enabled today on hubs (disabled on switches and routers) o Devices must be “polite” and follow these steps: 1. Listen” for traffic 2. “If another device is transmitting, wait to transmit 3. Otherwise, transmit (and keep listening) 4. If another device begins to transmit, stop and wait Controlled access, common in wireless LANs o 2 approaches: Access Request: Each device must get “permission” to transmit, similar to raising a hand (RTS/CTS) Polling Roll-call polling: Central device (controller) determines which devices can transmit Each client is checked periodically to see if it needs to transmit Hub Polling (token passing) One device begins the poll and then passes it to another device until it reaches them all 12 ITEC 3210 Notes York University Error Control with Focus on Ethernet By definition, Error control = Error prevention + Detection + Correction In general, network errors (include both corrupted and lost data) are caused by problems in transmission (not by humans) and should be prevented: Ethernet Error Detection Ethernet’s error detection scheme is Cyclic Redundancy Check algorithm o CRC on sending computer treats message as a single binary number, divides it by a pre-set number, uses remainder as CRC code (or check value) and send it in the frame o On receiving computer perform same calculations and compare the CRC code to that received: If they match, no error in transmission, otherwise an error is detected o Detection rate: >99.99999% for a 32-bit CRC code (CRC-32) used by Ethernet* Ethernet’s error correction: Not done in practice. Bad frames are discarded and the transport layer requests that data be re-sent (Ch.5) 13 ITEC 3210 Notes York University Data Link protocols with focus on Ethernet Each data link protocol has its message delineation, indicating the content of its different fields (see Fig. 4-8a) Examples of fields in the 802.3ac Ethernet frame above: o Preamble: Synchronizes clock of sender and receiver (overhead bits) o Start of the frame: Indicates the start of the frame (overhead bits) o Destination and Source MAC addresses (overhead bits) o (Frame) Data: 46B to 1500B (encapsulated IP packet + IP header) (information bits) o Frame Check Sequence (FCS): Contains the CRC code (overhead bits) Transmission efficiency: ∑ Information bits / (∑ Information bits + ∑ Overhead bits) If size of frame is 1500 B, including a total overhead is 33 B, Ethernet protocol Lecture 4 - Network Layer & Transport Layer Context Transport and Network layer with focus on TCP/IP o Key functions o Key protocols Addresses in IP networks o Types o Subnetting o Partially covered: Session management and protocols Network & Transport Layers 14 ITEC 3210 Notes York University Transport Layer o Links application and network layers; responsible for segmentation and reassembly o Session management; responsible for end-to-end delivery of messages Network Layer o Responsible for addressing and routing of messages Context and Applicability Transport layer applies to any IP network (but different applications may use different protocols; e.g., TCP, UDP, etc.) Network layer typically applies to traffic in MANs, WANs, Internet and some BNs, NOT within access and distribution layers o NOT applicable because within those layers traffic goes from the sending computer to the receiving computer via a hub, an AP or a switch using MAC addresses (Ch.7) o Applicable in some core-layer BNs and in all MANs/WANs/Internet, as traffic typically goes through at least one router. For BNs, other options exist (Ch.8) Context: Intro to Routers Types of routers* o “Wireless router”: NOT a router and NOT relevant here o Routers that connect an organizational network to MANs/WANs/Internet: Partially relevant (Ch 9) o Routers used for some BNs (option for core layers of large networks only): Relevant (Ch 8) o Routers used by MAN/WAN/Internet service providers to build their networks: Relevant Key functions of relevant routers o Find a path or the best path from a source to a destination (routing) o Move data on core layer or between MANs/WANs and on the Internet 15 ITEC 3210 o Notes York University Other key functions: Security (ACL) and protocol conversion (Ch.8, 10) Anatomy of Routers CPU(s) + memory (RAM, ROM, etc.) + special OS (like computers) Different types of ports, each has its own purpose (data communication, configuration, etc.) o Each data communication port (aka interface) can be configured to connect to one or several networks (Fig. 5-13 and 5-14) Processing based on: interfaces + routing protocols + routing tables + ACL Transport Layer: Functions & Protocols Divides on the sending computer messages into Protocol Data Units (PDU) called segments (in the case of TCP) or datagrams (in the case of UDP) and reassembles messages on the receiving computer(s) Provides links to the Application layer by including source and destination port number to identify application used (See Fig. 5-2 and 5.5) Delivers PDUs using a session management: o TCP for connection-oriented transmission (Reliable delivery: Handshake & Flow control); 20–24 bytes of header overhead o UDP for connectionless transmission (Fast but not reliable delivery: No handshake, No flow control & less overhead: 4–8 bytes) o Quality of Service (QoS) protocols or extensions of other protocols for special type of connection-oriented transmission which assigns PDUs of time-sensitive applications higher priority through a guaranteed data rate ➔Here, the selection of protocols (TCP, UDP or those supporting QoS) depends on the application used NOT the network designer. Typically UDP is used for supervisory applications and online games, TCP for most other useroriented applications. ➔Configuring QoS is done by the network administrator and requires QoScompliant devices and software to apply policies when needed (Ch. 12) Network Layer: functions & protocols Provides IP addressing (using IP) that identifies networks (using the Net ID portion of an IP address) and hosts1 (using the Host ID portion of the IP address) to deliver packets o Class A, Class B, and Class C IP addresses (see also CIDR ) o Public vs private IP addresses o Static vs dynamic addressing o Subnetting o Address resolution 1Host is a computer or device in a subnet (different from “host” in Ch.2) 16 ITEC 3210 Notes York University Provides routing by determining the path for PDUs (called packets) from sender to receiver: o Static versus dynamic routing o A few routing protocols (extra slides at the end) Network Layer: DHCP Dynamic addressing o Configuring each device manually is time consuming o Assigning addresses permanently can be inefficient when devices are not connected to network all the time o A server can supply IP addresses automatically o Dynamic Host Configuration Protocol (DHCP) Most common protocol for dynamic addressing Device sends out broadcast message DHCP server responds with IP settings Addresses are “leased” for a length of time (hours, days…) Network Layer: Address Resolution Host (server) name resolution o Translate host name to IP address o e.g., www.yorku.ca → 130.63.236.137 o Domain Name Service (DNS) MAC address resolution o Identify MAC address of the next device in the circuit o Address Resolution Protocol (ARP) in LANs Addresses in IP networks In IP networks, each computer has several addresses, each used by a different layer Application layer addresses (aka domain names): Assigned by software configuration o Typically assigned to servers NOT to clients o Public (assigned by Internet authorities) or private (assigned by network administrator) o Public address: Guaranteed to be globally unique + routable (global traffic can directly reach its location) o more readable than IP address numbers: www.yorku.ca vs. 130.63.236.137 Network layer (aka IP or Layer 3) addresses: Assigned by software configuration o Two versions: IPv4 (32-bit addresses) and IPv6 (128-bit addresses) 17 ITEC 3210 Notes York University o IPv4: For human readers the dotted decimal notation is used (4 bytes separated by dots) o Achievable range of IPv4 addresses is 0.0.0.0 through 255.255.255.255 but NOT ALL are valid o Byte allocation: Class A (1 B network + 3 B host), Class B (2 B + 2 B), Class C (3 B + 1 B) (Fig.5-9) o Public or private (private addresses require NAT in networks with Internet access) o Public addresses can be static (case of servers) or dynamic (case of clients with automated assignment) o Used by devices for routing Data link layer (aka MAC or physical or Layer 2) addresses: Assigned by hardware o 48-bit addresses represented using hexadecimal notation o Identify computers and devices, NOT subnets o Used by devices for (actual) delivery of frames Example of TCP/IP network Subnetting Address Classes (obsolete, but still used, esp. in LANs) 18 ITEC 3210 Notes York University Subnetting Basics Network [IP] address: o A particular (more details in next slide) range of IP addresses o Represented by a prefix followed by 0 or 0s for all the bits on the right (128.192.56.0) Objective of subnetting: o Create multiple “logical” networks where hosts in a same subnet have a same net ID and hosts from different subnets have different net IDs: Fig. 5-11 o Without subnetting, logical networks would require several Class A, B, or C network IP addresses each with its own unique net ID but no ISP will provide so many network IP addresses to a single organization Key benefits: o Better traffic management: Routers use net IDs to bring a packet closer to its destination (i.e., routing is based on the net ID portion of an IP address): Fig. 5-13 and 5-14 o No waste of IP addresses: A single network IP address is used for all logical networks needed (NOT one network IP address for each logical network) Now, typical problem to solve: planning network extension with a single network IP address o Meet requirement related to # of subnets needed then obtain the resulting # of hosts per subnet OR 19 ITEC 3210 o Notes York University Meet requirement related to # of hosts needed per subnet then obtain the resulting # of subnets Subnetting Rules and Example How: "Borrow" bits from the <Host ID> portion to create <Subnet ID> portion Rules and formulas: o At least 2 bits remain usable for the Host ID (What is the max # of bits one can borrow?) o Host bits cannot be all 1's or all 0's: (2h– 2) possible values where h is the # of bits used to represent hosts o Subnet bits: 2s possible values where s is the # of bits used to represent subnets o Subnet mask: Special IP address-like bit pattern used to extract the new net ID (p.124); can use slash notation (p.122) Your company has been assigned ONE Class C network IP address. Your task: Plan for network extension knowing that you will not have more than 20 hosts on any subnet. o Q1: How many bits must remain in the Host ID portion? o Q2: What is the maximum number of hosts in each subnet? o Q3:What is the maximum number of subnets possible? o Q4: What is the subnet mask required? Other considerations (time permitting): click here Routing Routing is the process of determining the path a packet takes to get to destination o Single path or many possible paths o Routers keep routing tables to make routing decisions Types of routing: o Centralized routing: Routing decisions made by one computer, NOT used anymore 20 ITEC 3210 o o Notes York University Static routing: Decisions are based on fixed routing tables developed by network admin ➔Used on simpler networks Dynamic routing: An initial routing table is built by network admin and updated automatically and periodically by routers based on exchanged information re. real-time traffic conditions. When multiple routes exist, dynamic routing find the best route based on metric(s) used by the protocol implemented by the router (see next slide) Dynamic Routing: Criteria & Protocols Criteria used for dynamic routing decisions Distance Vector uses the least number of hops (hop = # of routers from source to destination) Link State uses one or several metrics: capacity, response time, packet loss, cost, etc.. o ➔Preferred: Link State (can use several metrics & provides more reliable and upto-date paths) Choice of protocols to use depends on: o Application (Fig. 5-15 p.131): Inside autonomous systems (interior RP) vs. between autonomous systems (exterior RP) (more choices available for interior RP than for exterior RP) o Distance Vector, Link State or Hybrid and metric(s) used in the case of Link State or Hybrid Key interior routing protocols and other considerations Features of key interior routing protocols Routing Information Protocol v2 (RIPv2 ) o Dynamic distance vector protocol o OK for small networks (up to 4 routers) Old Open Shortest Path First (OSPF) o Dynamic link state protocol using as metric cost o Most widely used interior routing protocol on large enterprise networks o Fast Enhanced Interior Gateway Routing Protocol (EIGRP) o Popular dynamic hybrid link state /distance vector protocol developed by Cisco o Metrics used include transmission capacity, delay time, reliability and load for all paths o Faster than OSPF Lecture 5 - Wired and Wireless LANs (Ch 7*) Covered Major components of wired and wireless LANs 21 ITEC 3210 Notes York University Wired LANs : o The dominant standard: Ethernet (802.3x) o Operations: Topologies and media access control (MAC) o Types of Ethernet Wireless LANs o The dominant standard: 802.11x (NOT Ethernet-compliant but Ethernetcompatible) o Operations: Topologies and media access control (MAC) o Security, site survey and other design considerations Best practice design to improve LAN performance Exercises Notes o Inconsistencies in text: wireless LANs are NOT Ethernet LANs o Standard vs. manufacturer implementations: dual band APs + assumptions in Fig.7-9 o Backup slide: Examples of special-purpose LANs and their key features Not covered: LAN in SOHO environment (7.5.5) Wired and wireless LAN components Clients and servers (covered in Weeks 1 and 2) Network Interface Cards (NICs aka network adapters) : Provide MAC addresses to computers on wired and wireless LANs. Wireless LANs require wireless NICs Circuits : Carry signals o Wired LANs can use UTP or STP, or optical fiber cables, whereas wireless LANs use radio waves o Key characteristics and applications covered in Week 3 Connectivity devices used at the access layer (Fig.7-1, 7-2 and 7-3) o Hubs or switches: Connect computers via cables and act as repeater for incoming signals o Access point (AP): Use radio waves to connect wireless clients to a wired network (AP ≠ wireless router, see next slide) Operating Systems for servers (aka SOS or NOS): o Windows Server, Unix, Linux, etc. o Manages a server’s networking functions including traffic control and security (i.e., network profiles) Overlay network: Illustrations 22 ITEC 3210 Notes York University WIRED LANs Ethernet Topologies Topology refers to the layout of a network o Physical topology: How a network looks physically => Hub = Switch o Logical topology: How a network works conceptually (how data flow) => Hub ≠ Switch Hubs create a logical bus topology (aka shared Ethernet): o A hub is a half-duplex device that broadcasts frames it receives from one port to all other ports but only the computer with the MAC address matching that of the destination reads that frame o A hub is a Layer-1 device and creates a multipoint circuit (or collision domain) Switches create a logical star topology (aka switched Ethernet): o A switch is a full-duplex device that reads destination address of frames and sends them to the intended destination only using entries in its forwarding table o A switch is a Layer 2-device (sometimes also Layer-3) and creates point-topoint circuits o Three switching modes: Cut-through switching: Frames retransmitted as soon as destination address read (low latency but no error checking) Store-and-forward switching: Frames retransmitted after entire frame is received and error check is complete (higher latency) Fragment-free switching: Frames retransmitted once the header (first 64 bytes) is received and has no errors (compromise between cut-through and store-and-forward) => 23 ITEC 3210 Notes York University Most switches today use cut-through or fragment-free switching Ethernet’s Media Access Control Ethernet’s medium access control protocol is CSMA/CD o Carrier Sense: Devices listen before transmitting o Multiple Access: Multiple devices have network access o Collision Detection: If a collision is detected, the devices sending the frames wait a random amount of time before retransmitting the frame CSMA/CD is “on” on hubs and “off” on switches o Hubs create collision domains: think of street intersections o Switches eliminate (prevent) collisions: think of highway interchanges Ethernet’s Types Types of Ethernet: depend on device used, nominal data rates and type of cabling o Examples: See Fig. 7-6 and Fig.7-1 o 10BaseT (10 = 10 Mbps; T=twisted pair; base=baseband): obsolete o Fast Ethernet: Includes 100BaseTX and 100BaseF (F = fiber) o Gigabit Ethernet (GbE): 1000BaseT and 1000BaseF (10GbE, 40GbE and 100GbE used in BNs; 2.5 Gb and 5.0 Gb versions also exist) o 100BaseTX and 1000BaseT use Cat5/5e/6 o 100BaseF and 1000BaseF case use various [single-/multimode] fiber optic media and various [near-/short-wavelength IR; 770–1600 nm] wavelengths Standard practices o Switched Ethernet 100BaseTX and 1000BaseT are the most common forms of Ethernet o Combining different nominal data rates requires multi-speed devices (why?) o Security: Switches should be secured in locked closets Wireless LANs Wireless LANs basics Reminder: NOT an Ethernet standard (Fig. 7-7) but Ethernet-compatible (note four address fields, instead of Ethernet’s two) Key benefits: o Allows user mobility o Provides network connections where cabling is impossible or too expensive 24 ITEC 3210 Notes York University Clients, printers… Avoided for servers Topology: Physical star and logical bus A client and an AP must be tuned to the same channel to exchange data. For that, the client scans or listens to all available channels then selects the one the AP uses o Wireless LAN Media Access Control CSMA/CA (CA = collision avoidance) o Collision Avoidance: CSMA but clients wait a random period of time before sending o 2 types of access control: Distributed Coordination Function (DCF) vs Point Coordination Function (PCF) DCF: Receiver sends an ACK (or NAK) to sender for each frame (a frame not ACKed is retransmitted) in a shorter amount of time than the random period mentioned above. PCF: Access controlled by AP with RTS and CTS (solves the “hidden node” problem where clients can be too far apart to sense one is already transmitting). Less often used Wireless LANs standards Important considerations o Concurrent use of non-overlapping channels increases the capacity of the WLANs (!!) o Actual data rates drop based on # of users, range and interference (walls, furniture, etc.) o Dual band APs are needed if concurrently supporting both 802.11n @2.4GHz and 802.11ac 25 ITEC 3210 o o o Notes York University 802.11ad (60 GHz): current range 10 m (or less) is too small for organizational networks APs may require PoE switches for power If many APs are needed to cover a large area, an AP controller helps to centrally manage them (See figure) Wireless LANs security (Content different from text) Particularly important as wireless LANs are easy to access by unauthorized users Three security mechanisms (or access controls) Encryption: 4 successive protocols based on encryption keys + algorithm o Wired Equivalent Privacy (WEP): Encryption key is dynamically produced but relatively easy to crack (takes a couple of minutes) o Wi-Fi Protected Access (WPA): Improved encryption due to a key that is (i) longer and (ii) changed for every frame transmitted o WPA2 (aka 802.11i): Most popular method that improves encryption by using a stronger encryption algorithm (Advanced Encryption Standard or AES) o WPA3: Current standard whose enhancements include longer encryption keys, improved exchange of keys and stronger password requirements Authentication: Access control based on user ID, passwords and permissions MAC address filtering: Access control based on their MAC address (but MAC addresses can be easily spoofed) Site survey for wireless LANs Before installing a wireless LAN, a site survey helps determine: Feasibility of the coverage required (floor plan needed) Number of APs needed and specs for the AP (aka Wi-Fi) controller Potential sources of interference and AP locations Locations of switches and power sources AP frequency assignment (using non-overlapping channels) Need for a primary and a guest wireless LAN to prioritize traffic when APs are too busy Best Practices for LAN Performance 26 ITEC 3210 Notes York University One must locate potential bottlenecks (i.e., parts of the network slowing the flow of data), which are due to over-utilization of servers (80 % of CPU or more) or of circuit capacity, and both can be monitored on existing LANs or simulated (using NAD tools) on To-Be LANs o 1. Improving server performance (Supply-side of the equation) Select NOS with settings to fine-tune server performance (e.g., cache and memory management) Use servers with faster CPU, more RAM, faster read/write disks Add more servers to accommodate an increased demand on the network (other technical goal(s) involved?) o 2. Improving circuit capacity (Supply-side of the equation): Use wired switched LANs for primary network and wireless LANs as overlay networks for mobile users Cost permitting, select the fastest stable technology available; i.e., choose 1000BaseT (or F if required) over 100BaseT (or F if required) and 802.11ac over 802.11n Segment LANs properly with different switches for different access layers and proper assignment to LANs Use switches and APs with low latency o 3. Reducing demand on a network (Demand-side of the equation): Move appropriate files to clients using thick clients and micro-computing applications Shift users from wired LANs to wireless LANs or vice versa depending on which has the problem Encourage off-peak usage and conduct “lights out operations” (e.g. backups) in off-hours Use components to block unnecessary traffic on the network (e.g. no gaming, no video streaming…) Special-purpose LANs: Examples and related features Special-purpose LANs built in large organizations include: Data center: LAN that houses all primary servers and related networking devices (Fig. 72). Typical features include: o Server farms or clusters i.e., many servers performing the same task o Load balancer that “evenly” directs request or processing to each server in the cluster o Server virtualization that reduces costs by combining underutilized resources of physical servers into a new logical (or virtual) server to run applications Storage Area Networks (SAN): High speed and highest availability LAN devoted solely to data storage. It is different from a Network Attached Storage (NAS), another data 27 ITEC 3210 Notes York University storage solution consisting of an optimized server devoted solely to data storage and attached to a “regular” access layer (i.e., departmental LAN) E-Commerce edge: LAN that houses servers and devices that support e-commerce including, clusters of web servers connected to the Internet, application servers, database servers, storage servers, load balancer etc. Features include a high security requirement 28
0
You can add this document to your study collection(s)
Sign in Available only to authorized usersYou can add this document to your saved list
Sign in Available only to authorized users(For complaints, use another form )