ITEC 275 Computer Networks – Switching, Routing, and

WANs

Week 11Robert D’Andrea

Some slides provide by Priscilla Oppenheimer and used with permission

Agenda

• Learning Activities– Industry Tests– Build and Test a Prototype– Write and Implement a Test Plan– Tools for Testing a Network Design– Multicasting– QoS– Queuing and Traffic Shaping

Reasons to Test• Verify that the design meets key business and

technical goals• Validate LAN and WAN technology and device

selections• Verify that a service provider provides the agreed-

up service• Identify bottlenecks or connectivity problems• Determine optimization techniques that will be

necessary

Reasons to Test• Proving that your network design is better than a

competing design • Passing an “acceptance test” that gives you approval to

go forward with the network implementation • Convincing mangers and coworkers that your design is

effective• Identifying any risks that might impede

implementation and planning for contingencies• Determine how much additional testing might be

required. Will the new system be deployed as a pilot and undergo additional testing before being implemented

Testing Your Network Design• Use industry testing services• Build and test a prototype system• Use third-party and Cisco tools

Respected Independent Test Labs• The Interoperability Lab at the University of N

ew Hampshire (IOL)• ICSA Labs• Miercom Labs• AppLabs• The Tolly Group

Simple verses Complex Network Designs

• Simple network designs can rely on test results from vendors, independent labs, or trade journals to prove to your customer that your design will perform as intended.

• Complex network designs require more considerations. – Testing should be implemented in-house– Testing will require more that component testing.

There will be a need for unit, integration, and system testing.

Scope of a Prototype System• It’s not generally practical to implement a

full-scale system.• A prototype should verify important

capabilities and functions that might not perform adequately.

• Risky functions include complex, intricate functions and functions that were influenced by the need to make tradeoffs.

Live Production Network• Perform initial testing during off-hours to

minimize issues with user community, performance, and existing traffic flow

• Perform final testing during normal hours• Perform final testing at various times to

exercise network during typical loads

Components of a Test Plan

• Test objectives and acceptance criteria• The types of tests that will be run• Network equipment and other resources

required• Testing scripts• The timeline and milestones for the testing

project

Components of a Test Plan• Test objectives and acceptance criteria

Objectives and acceptance should be based on a customer’s business and technical goals

Acceptance of test results are acceptable by both the customer and the tester.

– Measure response time– Measure applications throughput– Measure the amount of time it takes to hear a dial

tone using Voice over IP– Establish a baseline measurement of CRC errors

Test Objectives and Acceptance Criteria

• Specific and concrete• Based on business and technical goals• Clear criteria for declaring that a test passed or

failed• Avoid biases and preconceived notions about

outcomes• If appropriate, reference a baseline

Types of Tests

• Application response-time tests• Throughput tests• Availability tests• Regression tests

Resources Needed for TestingTest plan should include a network topology drawing for tester to be able to reference.

A list of switches, routers, bridges, firewalls, servers, telephone equipment, and wireless access points.

A list of documented version numbers for hardware and software.

• Scheduled time in a lab either at your site or the customer’s site

• Power, air conditioning, rack space, and other physical resources

• Help from coworkers or customer staff• Help from users to test applications• Network addresses and names

Example Test Script

Network A Network B

Server 1

Firewall

Protocol Analyzer

Workstations

Protocol Analyzer

Example Test Script (continued)

• Test objective. Assess the firewall’s capability to block Application ABC traffic, during both light and moderately heavy load conditions.

• Acceptance criterion. The firewall should block the TCP SYN request from every workstation on Network A that attempts to set up an Application ABC session with Server 1 on Network B. The firewall should send each workstation a TCP RST (reset) packet.

Example Test Script (continued)1. Start capturing network traffic on the protocol analyzer on

Network A.

2. Start capturing network traffic on the protocol analyzer on Network B.

3. Run Application ABC on a workstation located on Network A and access Server 1 on Network B.

4. Stop capturing network traffic on the protocol analyzers.

5. Display data on Network A’s protocol analyzer and verify that the analyzer captured a TCP SYN packet from the workstation. Verify that the network layer destination address is Server 1 on Network B, and the destination port is port 1234 (the port number for Application ABC). Verify that the firewall responded to the workstation with a TCP RST packet.

Example Test Script (continued)6. Display data on Network B’s protocol analyzer and

verify that the analyzer did not capture any Application-ABC traffic from the workstation.

7. Log the results of the test in the project log file.

8. Save the protocol-analyzer trace files to the project trace-file directory.

9. Gradually increase the workload on the firewall, by increasing the number of workstations on Network A one at a time, until 50 workstations are running Application ABC and attempting to reach Server 1. Repeat steps 1 through 8 after each workstation is added to the test.

Example Test Script (continued)Host A sends a TCP SYNchronize packet to Host B

Host B receives A's SYN

Host B sends a SYNchronize-ACKnowledgement

Host A receives B's SYN-ACK

Host A sends ACKnowledge

Host B receives ACK.

TCP socket connection is ESTABLISHED.

- See more at: http://www.inetdaemon.com/tutorials/internet/tcp/3-way_handshake.shtml#sthash.R92nBBrG.dpuf

Tools for Testing a Network Design• Network-management and monitoring tools.

These monitoring tools are used to alert network management to problems and report significant network problems.

• Traffic generation tools• Modeling and simulation tools• QoS and service-level management tools• Protocol analyzer• http://www.topdownbook.com/tools.html

Protocol Analyzer Tool• A protocol analyzer is used to analyze traffic

behavior, errors, utilization, efficiency, and rates of broadcast and multicast packets.

A protocol analyzer can be a computer program or a piece of computer hardware that can intercept and log traffic passing over a digital network or part of a network. As data streams flow across the network, the sniffer captures each packet and, if needed, decodes the packet's raw data, showing the values of various fields in the packet, and analyzes its content according to the appropriate RFC or other specifications.

Simulation Tool• A simulation tool enables you to develop a model

of a network, estimate the performance of the network and compare alternatives for implementing the network.

iTrinegy Network Emulator (INE) products enable you to realistically recreate a wide variety of network conditions like latency, jitter, packet loss/error/reordering and bandwidth restrictions so that you can simulate environments such as Wide Area Networks (WANs), Wireless LANs, GPRS, 3G, IP over Radio / Radio over IP(RoIP), Satellite or MPLS networks.

Reasons to Optimize• Meet key business and technical goals• Use bandwidth efficiently• Control delay and jitter• Reduce serialization delay• Support preferential service for essential

applications• Meet Quality of Service (QoS) requirements

IP Multicast

IP Multicast Helps Optimize Bandwidth Usage

• With IP multicast, you can send a high-volume multimedia stream just once instead of once for each user

• Requires support for– Multicast addressing– Multicast registration (IGMP)– Multicast routing protocols

IP Multicast Addressing• Uses Class D multicast destination address

– 224.0.0.0 to 239.255.255.255

• Converted to a MAC-layer multicast destination address– The low-order 23 bits of the Class D address

become the low-order 23 bits of the MAC-layer address

– The top 9 bits of the Class D address are not used– The top 25 bits of the MAC-layer address are

0x01:00:5E followed by a binary 0

Internet Group Management Protocol (IGMP)

• Allows a host to join a multicast group• Host transmits a membership-report message

to inform routers on the segment that traffic for a group should be multicast to the host’s segment

• IGMPv2 has support for a router more quickly learning that the last host on a segment has left a group

Multicast Routing Protocols• Becoming obsolete

– Multicast OSPF (MOSPF)– Distance Vector Multicast Routing Protocol

(DVMRP)• Still used

– Protocol Independent Multicast (PIM)• Dense-Mode PIM• Sparse-Mode PIM

PIM (Protocol Independent Multicast)• Dense mode is used when there are many

members (employees listen to a company president). PIM is similar to DVMRP. Both use reverse-path forwarding (RPF) mechanism to compute the shortest (reserve) path between a source and all possible recipients of a multicast packet.

• Dense PIM does not require the computation of routing tables.

PIM (Protocol Independent Multicast)• Sparse mode utilizes a rendezvous point. A

rendezvous point provides a registration service for a multicast group.

• Sparse mode PIM relies on IGMP which let a host join a group by sending a membership-report message, and detach from a group by sending a leave message.

Serialization• Transmission delay or serialization delay

In a network based on packet switching, transmission delay (or store-and-forward delay) is the amount of time required to push all of the packet's bits into the wire. In other words, this is the delay caused by the data-rate of the link.

Transmission delay is a function of the packet's length and has nothing to do with the distance between the two nodes. This delay is proportional to the packet's length in bits.

Reducing Serialization Delay• Link-layer fragmentation and interleaving

– Breaks up and reassembles frames– Multilink PPP– Frame Relay FRF.12

• Compressed Real Time Protocol– RTP is used for voice and video– Compressed RTP compresses the RTP, UDP,

and IP header from 40 bytes to 2 to 4 bytes

A Few Technologies for Meeting QoS Requirements

• IETF controlled load service• IETF guaranteed service• IP precedence• IP differentiated services

IP Type of Service Field

• The type of service field in the IP header is divided into two subfields– The 3-bit precedence subfield supports eight

levels of priority– The 4-bit type of service subfield supports four

types of service

• Although IP precedence is still used, the type of service subfield was hardly ever used

IP Type of Service Field

Version Header Length

Type of Service Total Length

Identification Flags Fragment Offset

Time to Live Protocol Header Checksum

Source IP Address

Destination IP Address

Options Padding

0

Bit

8 15 2431

Precedence D T R C 0

Bit 0 3 4 5 6 7

Type of Service Subfield

D = DelayT = ThroughputR = ReliabilityC = Cost

IP Differentiated Services (DS) Field• RFC 2474 redefines the type of service field

as the Differentiated Services (DS) field– Bits 0 through 5 are the Differentiated Services

Codepoint (DSCP) subfield• Has essentially the same goal as the

precedence subfield• Influences queuing and packet dropping

decisions for IP packets at a router output interface

– Bits 6 and 7 are the Explicit Congestion Notification (ECN) subfield

IP Differentiated Services (DS) Field

Version Header Length

Differentiated Services Total Length

0 8 15 24 31

Differentiated Services Codepoint

0 6

Explicit Congestion Notification

Resource Reservation Protocol (RSVP)• RSVP complements the IP type-of-service,

precedence, DSCP, and traffic-class capabilities inherent in an IP header.

• RSVP supports mechanisms for hosts to specify QoS requirements for individual traffic flow.

• RSVP can be deployed on LANs and enterprise WANs to support multimedia applications or other types of applications with strict QoS requirements.

Resource Reservation Protocol (RSVP)• IP header type-of-service capabilities and

RSVP are examples of QoS signaling protocols.• In-band signaling means that bits within the

frame header signal to routers how the frame should be handled.

• Out-of-band signaling means that hosts send additional frames, beyond data frames, to indicate that a certain QoS is desired for a particular traffic flow.

Classifying LAN Traffic

• IEEE 802.1p• Classifies traffic at the data-link layer• Supports eight classes of service• A switch can have a separate queue for each

class and service the highest-priority queues first

Cisco Switching Techniques• Process switching is the slowest switching method• Fast switching allows highest throughput by

switching a packet using an entry in the fast-cache that was created when a previous packet to the same destination was processed.

• NetFlow switching is optimized for environments where services must be applied to packets to implement security, QoS features, and traffic accounting. Example Internet and enterprise network environment boundary.

Cisco Switching Techniques• Cisco Express Forwarding (CEF) is a

technique for switching packets quickly across large backbone networks and the Internet.

• CEF depended on a forwarding information base (FIB), rather than caching techniques.

• FIB allows CEF to use less CPU resources compared to other Layer 3 switching methods. FIB contains forwarding information for all routes in the routing tables.

Cisco Switching TechniquesWhy did CEF evolve?

With the introduction of web-based applications and other interactive applications that are characterized by sessions of short duration to multiple addresses.

It became very apparent that the cache-based system could not deliver the needed performance for these applications.

Cisco Queuing Services• First in, first out (FIFO) queuing store packets

when the network is congested and forward the packets in the order they arrived in when there is no congestion. Disadvantage: No packet priority scheme.

• Priority queuing ensures that important traffic is processed first. Priority is based on type of protocol, incoming interface, packet size, and source or destination address. The priorities are high, medium, normal, and low.

Cisco Queuing Services• Custom queuing is designed to allow the

network to be shared among applications with different minimum bandwidth or latency requirements. Custom queuing provides different amounts of queue space to different protocols and handles the queues in round-robin manner. A particular protocol can be prioritized by assigning it more queue space.

Custom queuing can be used to guarantee bandwidth at a potential congestion point.

Cisco Queuing Services• Custom queuing helps ensure that each traffic type

receives a fixed portion of available bandwidth and that when the link is under stress, no application receives more than a predetermined proportion of capacity.

• Weighted fair queuing (WFQ) operates from algorithms designed to reduce delay variability and provide predictable throughput and response time for traffic flows. Applications with small payloads are not starved of bandwidth by applications that send large packets.

Cisco Queuing Services• Class-based WFQ (CBWFQ) combines the best

scenarios of priority, custom, and weight-fair queuing.

Class-based WFQ allows you to define traffic classes based on matching criteria such as protocol, access control lists, and input interfaces.• Low latency queuing (LLQ) combines priority

queuing with CBWFQ. LLQ brings strict priority queuing to CBWFQ. Strict priority queuing allows delay-sensitive data such as voice to be sent before packets in other queues are sent.

Low-Latency Queuing

One queue always gets the green light

- Use this for voice

Combine this with class-based weighted fair queuing

- Define traffic classes based on protocols, access control lists, and

input interfaces

- Assign characteristics to classes such as bandwidth required and the

maximum number of packets that can be queued for the class

Priority Queuing

Packet in high queue?

NO

Dispatch Packet Continue

YESPacket in medium

queue?

NO

YESPacket in normal

queue?

NO

YESPacket in low

queue?

NO

YES

START

Custom Queuing

Reached transmission window size?

NODispatch PacketNext Queue

YES

START (with Queue 1)

Packet in Queue?

NO

YES

Low-Latency Queuing• One queue always gets the green light

– Use this for voice

• Combine this with class-based weighted fair queuing– Define traffic classes based on protocols,

access control lists, and input interfaces – Assign characteristics to classes such as

bandwidth required and the maximum number of packets that can be queued for the class

Random Early Detection (RED)• Congestion avoidance rather than congestion

management• Monitors traffic loads and randomly discards

packets if congestion increases• Source nodes detect dropped packets and slow

down– Works best with TCP

• Weighted Random Early Detection• Cisco’s implementation uses IP precedence or the DS

field instead of just randomly dropping packets

Traffic Shaping• Manage and control network traffic to avoid

bottlenecks• Avoid overwhelming a downstream router

or link• Reduce outbound traffic for a flow to a

configured bit rate– Queue bursts of traffic for that flow

Committed Access Rate (CAR)

• Cisco feature for classifying and policing traffic on an incoming interface

• Supports policies regarding how traffic that exceeds a certain bandwidth allocation should be handled

• Can drop a packet or change the IP precedence or DSCP bits

Summary

• An untested network design probably won’t work.

• It’s often not practical to test the entire design.• However, by using industry testing services

and tools, as well as your own testing scripts, you can (and should) test the complex, risky, and key components of a network design.

Summary

• Optimization provides the high bandwidth, low delay, and controlled jitter required by many critical business applications

• To minimize bandwidth utilization by multimedia applications, use IP multicast

• To reduce serialization delay, use link fragmentation and compressed RTP

• To support QoS and optimize performance, use IP precedence, DSCP, 802.1p. advanced switching and queuing methods, RED, CAR, etc.

Review Questions• Why is it important to test your network

design?• Why is regression testing important?• What are some characteristics of well-written

acceptance criteria?• What are some characteristics of a good

network simulation tool?

Review Questions• Why is it important to optimize your

network?• What has become of the IP type of service

field?• What are some methods for marking

packets to identify the need for priority handling?

• Compare and contrast Cisco queuing services.

This Week’s Outcomes

– Industry Tests– Build and Test a Prototype– Write and Implement a Test Plan– Tools for Testing a Network Design– Multicasting– QoS– Queuing and Traffic Shaping

Due this week

• 12-1 – Concept questions 9• 1-5-3 – Network design project

– New office network

Next week

• Review chapters 12 and 13 in Top-Down Network Design

• 13-1 – Concept questions 10• 4-2-3 – Networking practical experiences

– Basic network troubleshooting

Q & A

• Questions, comments, concerns?