Network Analysis and Design Introduction to Network Design

Network Design A network design is a blueprint for building a network The designer has to create the structure of the network [and] decide how to allocate resources and spend money

Elements of Good Network Design Deliver the services requested by users Deliver acceptable throughput and response times Cost efficiency Reliable Expandable Manageable Well-documented

Network Design Issues User requirements Locations of devices Characteristics of applications Types of traffic Topologies Routing protocols Budget Performance Etc.

Classifications of Network Design Build a new network Expand or upgrade the existing network Create the overlay network Virtual Private Network (VPN)

Types of Networks Access network: Backbone network: The ends or tails of networks that connect the small sites into the network LAN, campus network Backbone network: The network that connects major sites Corporate WAN

Objectives How to design a network using the correct techniques? Some common guidelines applicable for all types of network design

Top-Down Network Design Methodology A complete process that matches business needs to available technology to deliver a system that will maximize an organization’s success Don’t just start connecting the dots In the LAN, it is more than just buying a few devices In the WAN, it is more than just calling the phone company

Top-Down Network Design Methodology (Contd.) Analyze business and technical goals first Explore divisional and group structures to find out who the network serves and where they reside

Top-Down Network Design Methodology (Contd.) Determine what applications will run on the network and how those applications behave on a network Focus on applications, sessions, and data transport before the selection of routers, switches, and media that operate at the lower layers

Network Design Phases Requirement analysis Logical network design Physical network design

Phase I - Requirement Analysis Phase Analyze goals and constraints Characterize the existing network Characterize network traffic

Phase II - Logical Network Design Phase Map the requirements into the conceptual design Design a network topology Node locations Capacity assignment

Phase III - Physical Network Design Phase Select technologies and devices for your design Implementation

Business Goals Increase revenue Reduce operating costs Improve communications Shorten product development cycle Expand into worldwide markets Build partnerships with other companies Offer better customer support or new customer services

Recent Business Priorities Mobility Security Resiliency (fault tolerance) Business continuity after a disaster Networks must offer the low delay required for real-time applications such as VoIP

Business Constraints Budget Staffing Schedule Politics and policies

Information Goals of the project Scope of the project What problem are they trying to solve? How will new technology help them be more successful in their business? Scope of the project Small in scope: Allow sales people to access network via a VPN Large in scope: An entire redesign of an enterprise network Does the scope fit the budget, capabilities of staff and consultants, schedule?

Information (Contd.) Applications, protocols, and services Current logical and physical architecture Current performance

Technical Goals Scalability Availability Performance Security Manageability Usability Adaptability Affordability

Scalability Scalability refers to the ability to grow Network must adapt to increases in network usage and scope in the future Flat network designs don’t scale well Broadcast traffic affects the scalability of a network

Availability Availability is the amount of time a network is available to users Availability can be expressed as a percent up time per year, month, week, day, or hour, compared to the total time in that period 24/7 operation Network is up for 165 hours in the 168-hour week Availability is 98.21%

Availability (Contd.) Different applications may require different levels Some enterprises may want 99.999% or “Five Nines” availability

Availability (Contd.) An uptime of 99.70 % An uptime of 99.95 % Downtime = 0.003 x 60 x 24 x 7 30.24 mins per week An uptime of 99.95 % Downtime = 0.0005 x 60 x 24 x 7 5.04 mins per week An uptime of 99.999 % Downtime = 0.00001 x 60 x 24 x 365 5.256 mins per year

Availability (Contd.) System availability (R) is calculated from the component availability (Ri) Series: R =  Ri Parallel: R = 1 – (1 – Ri)

Availability (Contd.) R1 = 99.95%, R2 = 99.5% Series: Parallel: R = 0.9995 x 0.995 = 99.45% Decreases system availability Parallel: R = 1 – [(1 – 0.9995) x (1 – 0.995)] = 99.99975% Increases system availability

Availability (Contd.) 99.999% may require high redundancy (and cost) ISP 1 ISP 2 ISP 3 Enterprise

Availability (Contd.) Availability can also be expressed as a mean time between failure (MTBF), and mean time to repair (MTTR) Availability = MTBF / (MTBF + MTTR) A typical MTBF goal for a network that is highly relied upon is 4000 hours. A typical MTTR goal is 1 hour. 4000 / 4001 = 99.98% availability

Network Performance Common performance factors include Bandwidth Throughput Bandwidth utilization Offered load Accuracy Efficiency Delay (latency) and delay variation Response time

Bandwidth Vs. Throughput They are not the same thing Bandwidth is the data carrying capacity of a circuit Usually specified in bits per second Fixed Throughput is the quantity of error free data transmitted per unit of time Measured in bps, Bps, or packets per second (pps) Varied

Other Factors that Affect Throughput The size of packets Inter-frame gaps between packets Packets-per-second ratings of devices that forward packets Client speed (CPU, memory, and HD access speeds) Server speed (CPU, memory, and HD access speeds) Network design Protocols Distance Errors Time of day etc.

Throughput of Devices The maximum PPS rate at which the device can forward packets without dropping any packets Theoretical maximum is calculated by dividing bandwidth by frame size, including any headers, preambles, and interframe gaps

Throughput of Devices (Contd.) Frame Size (Bytes) Theoretical Max PPS (100-Mbps Ethernet) 64 148,800 128 84,450 256 45,280 512 23,490 768 15,860 1024 11,970 1280 9,610 1518 8,120

Bandwidth, Throughput, Load 100 % of Capacity Throughput Actual Ideal 100 % of Capacity Offered Load

Throughput Vs. Goodput Most end users are concerned about the throughput for applications Goodput is a measurement of good and relevant application layer data transmitted per unit of time In that case, you have to consider that bandwidth is being “wasted” by the headers in every packet

Utilization The percent of total available capacity in use For WANs, optimum average network utilization is about 70% For hub-based Ethernet LANs, utilization should not exceed 37%, beyond this limit, collision becomes excessive

Utilization (Contd.) For full-duplex Ethernet LANs, a point-to-point Ethernet link supports simultaneous transmitting and receiving Theoretically, Fast Ethernet means 200 Mbps available Gigabit Ethernet means 2 Gbps available 100% of this bandwidth can be utilized Full-duplex Ethernet is becoming the standard method for connecting servers, switches, and even end users' machines

Efficiency Large headers are one cause for inefficiency How much overhead is required to deliver an amount of data? How large can packets be? Larger better for efficiency (and goodput) But too large means too much data is lost if a packet is damaged How many packets can be sent in one bunch without an acknowledgment?

Efficiency (Contd.) Small Frames (Less Efficient) Large Frames (More Efficient)

Delay from the User’s Point of View Response Time The time between a request for some service and a response to the request The network performance goal that users care about most A function of the application and the equipment the application is running on, not just the network Most users expect to see something on the screen in 100 to 200 ms The 100-ms threshold is often used as a timer value for protocols that offer reliable transport of data

Delay from the Engineer’s Point of View Propagation delay Signal travels in a cable at about 2/3 the speed of light in a vacuum Relevant for all data transmission technologies, but especially for satellite links and long terrestrial cables Geostationary satellites: propagation delay is about 270 ms for an intercontinental satellite hop Terrestrial cables: propagation delay is about 1 ms for every 200 km

Delay from the Engineer’s Point of View (Contd.) Transmission delay Also known as serialization delay Time to put digital data onto a transmission line Depends on the data volume and the data rate of the line It takes about 5 ms to output a 1,024 byte packet on a 1.544 Mbps T1 line

Delay from the Engineer’s Point of View (Contd.) Packet-switching delay The latency accrued when switches and routers forward data The latency depends on the speed of the internal circuitry and CPU the switching architecture of the internetworking device the type of RAM that the device uses Routers tend to introduce more latency than switches QoS, NAT, filtering, and policies introduce delay

Delay from the Engineer’s Point of View (Contd.) Queueing delay The average number of packets in a queue on a packet-switching device increases exponentially as utilization increases

Queuing Delay and Bandwidth Utilization Number of packets in a queue increases exponentially as utilization increases

Delay Variation (Jitter) The amount of time average delay varies Users of interactive applications expect minimal delay in receiving feedback from the network Users of multimedia applications require a minimal variation in the amount of delay Delay must be constant for voice and video applications Variations in delay cause disruptions in voice quality and jumpiness in video streams

Delay Variation (Jitter) (Contd.) Short fixed-length cells, for example ATM 53-byte cells, are inherently better for meeting delay and delay-variance goals Packet size tradeoffs Efficiency for high-volume applications versus low and non-varying delay for multimedia

Delay Variation (Jitter) (Contd.) Audio/video applications minimize jitter by providing a buffer that the network puts data into Display software or hardware pulls data from the buffer

Accuracy Data received at the destination must be the same as the data sent by the source Error fames must be retransmitted, which has a negative effect on throughput In IP networks, TCP provides retransmission of data For WAN links, accuracy goals can be specified as a bit error rate (BER) threshold Fiber-optic links: about 1 in 1011 Copper links: about 1 in 106

Accuracy (Contd.) On shared Ethernet, errors often result from collisions Collisions happen in the 8-byte preamble of the frames (not counted) Collisions happen past the preamble and somewhere in the first 64 bytes of the data frame (legal collision) Collisions happen beyond the first 64 bytes of a frame (late collision)

Accuracy (Contd.) Late collisions are illegal and should never happen (too large network) A goal for Ethernet collisions: less than 0.1% affected by a legal collision Collisions should never occur on full-duplex Ethernet links In wireless LAN 802.11 CSMA/CA, collisions can still occur

Security Security design is one of the most important aspects of enterprise network design Security problems should not disrupt the company's ability to conduct business The cost to implement security should not exceed the cost to recover from security incidents

Security (Contd.) Network Assets Hardware Software Applications Data Intellectual property Trade secrets Company’s reputation

Affordability Affordability is sometimes called cost-effectiveness A network should carry the maximum amount of traffic for a given financial cost Financial costs include nonrecurring equipment costs and recurring network operation costs Campus networks: low cost is often more important than availability and performance. Enterprise networks: availability is usually more important than low cost

Affordability (Contd.) Monthly charges for WAN circuits are the most expensive aspect of running a large network How to save Use a routing protocol that minimizes WAN traffic Improve efficiency on WAN circuits by using such features as compression Eliminate underutilized trunks Use technologies that support oversubscription

Adaptability Avoid incorporating any design elements that would make it hard to implement new technologies in the future Change can come in the form of new protocols, new business practices, new traffic patterns

Usability The ease of use with which network users can access the network and services Usability might also include a need for mobility Some design decisions will have a negative affect on usability: Strict security, for example

Characterizing a Network (Why?) Verify that a customer's technical design goals are realistic Understand the current topology Locate existing network segments and equipment Locate where new equipment will go Develop a baseline of current performance

Characterizing a Network (What?) Infrastructure Addressing and naming Wiring and media Architectural and environmental constraints Health

Infrastructure Develop a set of network maps Learn the location of major internetworking devices and network segments

Infrastructure (Contd.) Information to collect Geographical locations LAN, WAN connections Buildings and floors, and possibly rooms Location of major servers or server farms Location of routers and switches Location of mainframes Location of major network-management stations Location and reach of virtual LANs (VLANs) Etc.

Infrastructure (Contd.) Medford Fast Ethernet 50 users Roseburg Fast Ethernet 30 users Frame Relay CIR = 56 Kbps DLCI = 5 Frame Relay CIR = 56 Kbps DLCI = 4 Gigabit Ethernet Grants Pass HQ 16 Mbps Token Ring Grants Pass HQ Fast Ethernet 75 users FEP (Front End Processor) IBM Mainframe T1 Web/FTP server Eugene Ethernet 20 users T1 Internet

Addressing and Naming IP addressing for major devices, client networks, server networks What to consider? Private/public address Classless/classful addressing Variable-length subnet mask (VLSM) Route aggregation or supernetting Discontiguous subnets

Discontiguous Subnets Area 0 Network 192.168.49.0 Router A Router B Area 1 Subnets 10.108.16.0 - 10.108.31.0 Area 2 Subnets 10.108.32.0 - 10.108.47.0

Wiring and Media Document the types of cabling in use as well as cable distances Distance information is useful when selecting data link layer technologies based on distance restrictions

Wiring and Media (Contd.) Single-mode (SM) fiber Multi-mode (MM) fiber Shielded twisted pair (STP) copper Unshielded-twisted-pair (UTP) copper Coaxial cable Microwave Laser Radio Infra-red

Architectural Constraints Make sure the following are sufficient Air conditioning Heating Ventilation Power Protection from electromagnetic interference

Architectural Constraints (Contd.) Make sure there’s space for: Cabling conduits Patch panels Equipment racks Work areas for installing and troubleshooting equipment

Wireless Installations Reflection Signal bounces back and interferes with itself Metal surfaces such as steel girders, scaffolding, shelving units, steel pillars, and metal doors Implementing a WLAN across a parking lot can be tricky because of metal cars that come and go

Wireless Installations (Contd.) Absorption Energy of the signal can be absorbed by the material in objects through which it passes Reduces signal level Water has significant absorption properties, and objects such as trees or thick wooden structures can have a high water content Implementing a WLAN in a coffee shop can be tricky if there are large canisters of liquid coffee

Wireless Installations (Contd.) Refraction RF signal is bent when it passes from a medium with one density into a medium with another density The signal changes direction and may interfere with the nonrefracted signal It can take a different path and encounter other, unexpected obstructions, and arrive at recipients damaged or later than expected

Wireless Installations (Contd.) Diffraction Similar to refraction Like refraction, the signal is bent around the edge of the diffractive region and can then interfere with that part of the signal that is not bent

Wireless Installations (Contd.) Boost the power level to compensate for variable environmental factors The additional power added to a transmission is called the fade margin

Health Performance Availability Bandwidth utilization Accuracy Efficiency Response time Status of major routers, switches, and firewalls

Develop a Performance Baseline How much better the new internetwork performs once your design is implemented Baseline of normal performance should not include nontypical problems caused by exceptionally large traffic loads The decision whether to measure normal performance, performance during peak load, or both, depends on the goals of the network design

Characterize Availability Date and Duration of Last Major Downtime Cause of Last Major Downtime MTBF MTTR Enterprise Segment 1 Segment 2 Segment n

Utilization Measurement of how much bandwidth is in use during a specific time interval Different tools use different averaging windows for computing network utilization Trade-off between amount of statistical data that must be analyzed and granularity

Utilization in Minute Intervals

Utilization in Hour Intervals

Utilization (Contd.) The size of the averaging window depends on your goals When troubleshooting network problems, keep the interval very small, either minutes or seconds For performance analysis and baselining purposes, use an interval of 1 to 5 minutes For long-term load analysis, to determine peak hours, days, or months, set the interval to 10 minutes

Bandwidth Utilization by Protocol Relative Network Utilization Absolute Network Utilization Broadcast Rate Multicast Rate Protocol 1 Protocol 2 Protocol 3 Protocol n

Accuracy Bit error rate (BER) Frame error rate (FER) Packet loss Collision Runt (partial) frame Healthy network should not have more than one bad frame per megabyte of data

Characterize Packet Sizes Increasing the maximum transmission unit (MTU) on router interfaces can also improve efficiency Increasing MTU can increase serialization delay

Characterize Packet Sizes (Contd.)

Characterize Packet Sizes (Contd.) Small frames consist of control information and acknowledgments Data frames fall into the large frame-size categories Frame sizes typically fall into what is called a bimodal distribution

Characterize Response Time A more common way to measure response time is to send ping packets and measure the round-trip time (RTT) Variance measurements are important for applications that cannot tolerate much jitter You can also document any loss of packets

Characterize Response Time (Contd.) Node A Node B Node C Node D Node A Node B Node C Node D X X X X node = router, server, client, or mainframe

Checking Status of Major Devices CPU utilization How many packets it has processed How many packets it has dropped Status of buffers and queues You can use SNMP or commands in the devices

Characterizing Network Traffic (Why?) Analyze network traffic patterns to help you select appropriate logical and physical network design solutions to meet a customer's goals

Network Traffic Factors Location of traffic sources and sinks Traffic load Traffic behavior

Traffic Flow Information transmitted between communicating entities during a single session Flow attributes: addresses for each end of the flow direction symmetry path number of packets or bytes

Traffic Flow Types Terminal/host Client/server Peer-to-peer Server/server Voice over IP

Terminal / Host Examples: Telnet, ssh Usually asymmetric: terminal sends a few characters and the host sends many characters In some full-screen terminal applications, the terminal sends characters typed by the user and the host returns data to repaint the screen The screen is usually 80 characters wide by 24 lines long, which equals 1920 characters The full transfer is a few thousand bytes

Client / Server Examples: FTP, HTTP Usually bidirectional and asymmetric Requests are typically small frames except when writing data to the server Responses range from 64 bytes to 1500 bytes or more, depending on the MTU of the data link layer

Peer-to-Peer Examples: Workgroup, videoconferencing, P2Ps No hierarchy and no dedicated server Usually bidirectional and symmetrical Another example is a meeting between business people at remote sites using videoconferencing equipment Information dissemination in a class is a client/server model

Server / Server To implement directory services, to cache heavily used data, to mirror data for load balancing and redundancy, to back up data, and to broadcast service availability Generally bidirectional With most server/server applications, the flow is symmetrical, but in some cases there is a hierarchy of servers, with some servers sending and storing more data than others

VoIP The flow associated with transmitting the audio voice is separate from the flows associated with call control The voice flow for transmitting the digital voice is essentially peer-to-peer The call control flow for call setup and teardown is a client/server flow

Traffic Load Network capacity is sufficient to avoid bottleneck Key parameters: Number of stations Average time that a station is idle between sending frames Time required to transmit a message once medium access is gained Application usage patterns

Traffic Load (Contd.) Traffic load caused by applications Terminal screen: 4 Kbytes Simple e-mail: 10 Kbytes Simple web page: 50 Kbytes High-quality image: 50,000 Kbytes Database backup: 1,000,000 Kbytes or more

Traffic Load (Contd.) Protocol overhead IPX: 30 bytes TCP: 20 bytes IP: 20 bytes Ethernet: 18 + 8-byte preamble + 12-byte interframe gap (IFG) HDLC: 10 bytes

Traffic Behavior Broadcast Goes to all network stations on a LAN All ones data-link layer destination address FF: FF: FF: FF: FF: FF Doesn’t necessarily use huge amounts of bandwidth But does disturb every CPU in the broadcast domain

Traffic Behavior (Contd.) Multicast Goes to a subset of stations 01:00:0C:CC:CC:CC (Cisco Discovery Protocol) Should just disturb NICs that registered to receive it Requires multicast routing protocol on internetworks

Traffic Behavior (Contd.) Broadcast/multicast traffic is necessary and unavoidable share topology information advertise services locate services addresses and names No more than 20% of the network traffic, otherwise segment the network using routers or VLANs

Traffic Behavior (Contd.) Layer 2 devices, such as switches and bridges, forward broadcast and multicast frames out all ports Router does not forward broadcasts or multicasts All devices on one side of a router are considered part of a broadcast domain VLANs can also limit the size of a broadcast domain based on membership