Lecture04.pdf

Lecture 4: Introduction to Computer Network Design Instructor: Hussein Al Osman Based on Slides by: Prof. Shervin Shirmohammadi Hussein Al Osman CEG4190 4-1  Compute...

Lecture 4: Introduction to Computer Network Design Instructor: Hussein Al Osman Based on Slides by: Prof. Shervin Shirmohammadi Hussein Al Osman CEG4190 4-1  Computer Networks Hussein Al Osman CEG4190 4-2  Background Network design has generally been considered an art consisting of a set of rules acquired through experience. – For example the 80/20 rule where 80% of a network’s traffic is local and 20% is remote. – “Bridge when you can, route when you must”. Based on an antiquated view that bridging was easier and cheaper. Network technologies and requirements have increased tremendously. There are many options available now. Hussein Al Osman CEG4190 4-3  Factors that affect Network Design Capacity planning (a major factor in the design of the network). – Generally the solution to a problem will be “throw bandwidth to the problem”. Sometimes and especially recently it offers some relief but it is does not solve all problems. Delay performance – Has become very important in real-time services and needs to be optimized. Network Reliability, Maintainability, and Availability (RMA). – A factor that is really important for customer satisfaction. Hussein Al Osman CEG4190 4-4  The Technical View A “Network” really can be thought of as of three things and they all need to be considered when working on a network design project – Connections – Communications – Services Connections: provided by Hardware that ties things together – Wire/Fiber Transport Mechanisms – Routers – Switches/Hubs – Computers Hussein Al Osman CEG4190 4-5  The Technical View (…) Communications: provided by Software – A common language for 2 systems to communicate with each other. – Interoperability Protocols TCP/IP (Internet/Windows NT), IPX / SPX (Novell Netware 4), AppleTalk, etc. Services: the Heart of Networking - specially important today: the Internet in particular is moving quickly from a connection emphasis to a service oriented network. Cooperation between 2 or more systems to perform some function – Applications driven – telnet, ftp, http, SNMP, UDP, etc. Hussein Al Osman CEG4190 4-6  High Level View of Network Design Section 1 Section 2 Section 3 (Conceptual Model) (Logical Model) (Physical Model) Analysis Architecture Design Requirements, Relationships within and Flows, Risks between Network Technology, Equipment Functions Choices, Connectivity Choices Hussein Al Osman CEG4190 4-7  Overview of Network Analysis State of existing network Problems with existing network Network goals Requirements from users, applications, devices Network Analysis Description for requirements for the network Descriptions of traffic flows Mappings of applications and devices to network Hussein Al Osman CEG4190 4-8  Overview of Network Architecture Description for requirements for network Descriptions of traffic flows Mappings of applications and devices to network Network Architecture Reference architecture for network (end to end structure). Relationships between functions (routing, management, performance, security). Descriptions of interactions, trade-offs, dependencies, and constraints. Hussein Al Osman CEG4190 4-9  Overview of Network Design Reference architecture for network Relationships between network functions Descriptions of interactions, trade-offs, dependencies, and constraints. Network Design Physical details, evaluate and choose technologies. Strategies for interoperability. Evaluation criteria. Hussein Al Osman CEG4190 4-10  Hierarchy Levels of Hierarchy Levels of Diversity Core Access … … … … … … … … … End Users Hussein Al Osman CEG4190 4-11  Network - 1 Level Broadcast Domain Flat Bridged or Switched Network: what is the problem with this? Hussein Al Osman CEG4190 4-12  Network - 2 Levels Broadcast Domain Broadcast Domain Broadcast Domain What is the hierarchy degree? Hussein Al Osman CEG4190 4-13  The Need for Diversity Flows are forced through hierarchy Network Network Network Network Network Network Network Hussein Al Osman CEG4190 4-14  Flows Improvement with a CDN Network Network Network Network Network Network Network CDN Content Delivery Network provides direct connectivity Diversity (aka, Interconnectivity, aka Redundancy): interconnecting the network at different levels to achieve desired performance. Hussein Al Osman CEG4190 4-15  Network & System Complexity 4th Generation – Decision-making Complexity Interactions 3rd Generation – Services Interactions 2nd Generation – Interoperability 1st Generation – Connectivity Where are we today? Hussein Al Osman CEG4190 4-16  Architecture & Design Defensibility Very important and often overlooked part of network analysis and design. – During analysis you are gathering data and making decisions. – Details about how one got to those decisions is generally lost. – It is important to keep an audit trail. A set of documents, data, and decisions for the architecture and design. – This audit would answer questions like: Why did you choose this technology? Why does it cost this amount? Time spent at this phase can save large amount of time and resources later. What type of information? (Time stamp it) – Requirements (What type of Network Requirements would you expect?) – Problem definitions (There is no end to these). – Goals, Decisions – Data (What type of data?) Hussein Al Osman CEG4190 4-17  A Systems Methodology The idea is to view the network and a subset of its environment (what it impacts or interacts with) as a system. – This primarily implies looking at the services associated with the network (3rd generation networks). – Interactions and dependencies between the network its users, applications, and devices arises from this analysis. Traditional Device Device view of a system Network Better generic components of a User User system Application Application Device Device Network Hussein Al Osman CEG4190 4-18  A Systems Methodology (granularity) Components help identify interfaces User User User-Application Interface (Displays, UI) Application Application Application-Device Interface (API, QoS) Device Device Device-Network Interface (Device Drivers) Network  Comparison to OSI Application User Presentation Session Application Transport Network Device Data Link Physical Network OSI System Levels Hussein Al Osman CEG4190 4-20  Service Characteristics Service Characteristics: individual network performance and functional parameters that are used to describe the services. – These can be considered as requirements for the network. – E.g.: Providing 1.5 Mb/s peak capacity to a user Guaranteeing a maximum RTT of 100 msec Defining a specific security level for a group of users User-server WAN WAN service LAN-WAN Hussein Al Osman CEG4190 4-21  Service Metrics Services must also be configurable, measureable, and verifiable within the system to ensure that applications, users and devices are getting the services requested – Leads to accounting and billing How: using service metrics. Service Metrics: measurements of characteristics in the actual network to monitor, verify, and manage services. Services must be described and provisioned end-to-end at all network components between well-defined demarcation points. – Otherwise some components might not be capable of supporting the services. Hussein Al Osman CEG4190 4-22  Service Levels Service Levels: grouping of a number of service characteristics. E.g., combining capacity (1.5 Mb/s) and reliability (99.99% uptime). Can be offered by service providers to customers, in packages: – Basic (No Priority) – Gold (High Capacity) – Platinum (High Capacity, Reliability, Low Delay) Hussein Al Osman CEG4190 4-23  Grouping Characteristics into Services Item Examples Characteristics Delay (100 ms) Capacity (10 Mb/s) RMA (99.999% Uptime) Security (Encryption) Service Metrics End-to-End delay, Round-Trip delay, capacity, throughput, Buffer/Queue Utilization, priority levels. Service Levels Basic (No Priority) Gold (High Capacity) Platinum (High Capacity, Reliability, Low Delay) System Components & Network Services (1) GigE GigE OC-48 OC-48 GigE FE POS/O C-48 GigE User PCs Servers GigE Router Router Switch (100) (4) Switch Possible Demarcation Points 10 Potential aggregate capacity (Gb/s) 0.1 1 Distance along transmission path Hussein Al Osman CEG4190 4-25 System Components & Network Services (2) GigE GigE OC-48 OC-48 GigE FE FE FE POS/O C-48 Servers GigE Router Router GigE Firewall FE User Switch Switch Switch PCs (4) (100) Potential aggregate 10 capacity (Gb/s) 0.1 1 Distance along transmission path Hussein Al Osman CEG4190 4-26  Types of Network Service Best-effort service – No control how the network will satisfy the service request. Unpredictable and unreliable service with variable performance. Guaranteed service – Opposite of best-effort service. Predictable and reliable. Implies a contract between the user and the service provider. Predictable service – Some degree of predictability without accountability. Service requirements must be configurable, measurable, and verifiable. Control of the system is required. Hussein Al Osman CEG4190 4-27  Best Effort Performance Fast Ethernet Resource Loading 100 Capacity Fast Ethernet (Mb/s) Congestion begins Capacity to load network Available Throughput 50 0 Number of Concurrent Traffic Flows Hussein Al Osman CEG4190 4-28  Guaranteed Performance Telephony network, with Call Admission Control (CAC) Resource System Resources are 100 fully loaded at this Capacity Fast Ethernet (Mb/s) Loading point No more calls allowed until resource 50 are available As calls are added resource loading increases 0 Number of Concurrent Traffic Flows (Calls) Hussein Al Osman CEG4190 4-29  Best Effort vs. Guaranteed Best Effort: allows maximum number of traffic flows, but performance degradation likely. Guaranteed: performance will be met, but fewer traffic flows can be accepted. Many applications need a hybrid of the two (Predictable Effort). – E.g., VoIP requires characteristics of CAC while operating on top of best-effort Internet. Hussein Al Osman CEG4190 4-30  Performance Characteristics Capacity – is a measure of the system’s ability to transfer information (voice, data, video, etc). Delay – is a measure of the time difference in the transmission of information across the system. RMA – refers to Reliability, Maintainability, and Availability Hussein Al Osman CEG4190 4-31  Capacity Bandwidth: theoretical capacity of one or more network devices or communication links in the system. – Raw capacity does not take into account overhead from higher-layer protocols or performance loss due to device inefficiencies. Throughput: is the realizable capacity of the system or its network devices. E.g., theoretical SONET OC-3c link capacity (bandwidth) is 155.52 Mb/s – This does not include data-link, network, or transport-layer protocol overhead. – Performing at line rate when T=155.52 Mb/s In reality, when tested with TCP, SONET OC-3c link capacity (throughput) is 80 to 128 Mb/s – Where does the majority of this overhead coming from? Hussein Al Osman CEG4190 4-32  Delay Sources: – Propagation, transmission, queuing, and processing There are several ways to measure delay – Direction (end-to-end, round-trip) – Latency – time it takes to process information and reply. E.g., response time of a network device. – Jitter or delay variation – change in delay over time. Real- time and near real-time require strict delay variation. Jitter is NOT delay Hussein Al Osman CEG4190 4-33  RMA Reliability: a statistical indicator of the frequency of failure of the network. – Represents un-scheduled outages of services. – Predictable behavior – delivery of information must occur within well- known time boundaries. Maintainability: a statistical measure of the time to restore the system to fully operational status after it has experienced a fault. – It is generally expressed as a mean-time-to-repair (MTTR). Availability: relationship between the frequency of mission- critical failures and the time to restore the service. – Defined as the mean time between failures divided by the sum of mean time to repair and mean time between failures: A = (MTBF) / (MTBF + MTTR) Hussein Al Osman CEG4190 4-34  Thank You! Hussein Al Osman CEG 4190 3-35