Requirement Analysis Guidelines. 2 Process Model for Requirement Analysis Gather Requirements Develop Service Metrics To measure performance Characterizing.

Requirement Analysis

Guidelines

2

Process Model for Requirement Analysis

Gather Requirements

Develop Service MetricsTo measure performance

Characterizing Behavior

Develop PerformanceThresholds

Distinguishing betweenService Requirements

User/ApplicationPerf. Modifiers

Network Mgmt.Variables

DevelopApplications Map

Initialconditions

ApplicationTypes/groups

Guidelines inDistinguishingservices

3

Initial Condition

• Business _______• _______ of design

– New design, modification of an existing design, outsourcing

• _________ of design– Network size, distance, number of sites

• __________– Political, administrative, financial.

4

Importance of Initial Condition

• Knowing the type and scope of the network project will help you to________your design efforts.

• These conditions apply when you are doing the _________ of a design.

• Stand a much better chance of understanding and working with these ______________.

5

Step 1. Requirements collection

• Develop a ______ or questionnaire for users.

• Arrange a meeting or whiteboard session to discuss their _____________.

• Spend time with one or more high-performance _______ to get immersed in their environment.

6

Results of Requirement Collection

• A listing of __________– A concise, easy-to-use list

– Help you to identify what’s _________ for your environment.

– Determine __________s between various requirements.

• Mapping Applications– Application _________ information used to show their

relative locations in the design environment.

7

Step 2: Developing service metrics to measure performance

• Service metrics are either actual ________ variables in the network, or are derived from measured variables.

• Performance requirement is useful only if it is _________, __________, and _______ within the network.

• Service metrics will depend on your design environment and the types of equipment you implement in the network.

8

Service metrics for reliability

• Availability– In terms of _______ uptime and downtime.

• Recoverability or stability– In terms of mean time between failure (______), mean

time between service outage (_________), and mean time to repair (________)

• Error and loss rates at various levels– Bit error rate (_____), ___________ (CLR), cell

misinsertion ratio (CMR), and frame and packet loss rates.

9

Service Metrics for Capacity

• Data _______s– In terms of peak data rate (PDR), sustained data

rate (SDR), and minimum data rate.

• Data ______– Including burst size and duration

10

Service Metrics for Delay

• End-to-end, ___________, or system delay

• Latency

• Delay ________

• Timeliness.

11

Results of Service Metrics

• Configurable and measurable quantities in network _____________ systems, such as SNMP, CMIP, ..

• Other available tools– Ping– Pathchar– Tracert

Service metric where metric will be measurement measured in system method

12

Wide Area Network

LAN LAN

Network Monitorstation

SNMP/CMIP is used to get packet loss data

Ping is used between variousInterfaces to monitor delaysIn the network.

13

Step 3: Characterizing Behavior

• Usage ________s– number of sessions/user-day– Average time/session– Average number of active user sessions

• Application __________– Data size to be processed– Frequency and time duration for data to be passed across the network– __________ characteristics (direction)– Degree of multicasting

• Models– Merge to one session– Statistical formulas for modeling

14

Results of Characterization Behavior

• Apply a ____________ to the performance requirement for each application.

active

active

active

active active

active active active active

active active

Time

Session 1

Session 2 Session 3

Number of simultaneous sessions

Frequency

15

Step 4: Developing Performance Thresholds

• _______ for distinguishing between services– A_________ threshold for separating performance

requirements into low-performance and high-performance.

– An _____________________ threshold for separating requirements into low-performance and high-performance.

– _______ services will have known bounds or guarantees.

16

Step 4: Developing Performance Thresholds

• Criteria for distinguishing between services– A general threshold– An environment-specific threshold – Specific services

17

Reliability Requirements

• Availability– How much downtime can be expected per

_____________

Availability Amount of allowed downtime (hours [h], minutes[m](% uptime) seconds [s] per time period)

Yearly monthly weekly Daily

95% 438 h 36.5 h 8.4 h 1.2 h

18

How can availability be measured?

• _________ should availability be measured ?

• What service ________ can be used to measure it?

How we measure availability indicates what we consider to be _________ in the system.

19

Guidelines for Availability Measurement

• Availability is measured __________ between– Users, hosts, or networks, or– Particular users, hosts, or networks.

• A loss of availability in ____ part of the system is counted against any overall availability metrics.

• General threshold is ______ %

20

Availability as Measured End-to-End between Networks

Wide Area Network

LAN LAN

Network Monitors

NetworkInterfaces

21

Availability as Measured Selectively Between Networks

Server LAN

User LANAvailability

Availability

22

Recoverability• Using MTBF/MTBSO and MTTR.

– _______: estimation of how long system outages may last; related to repair time in a service contract.

– MTBF/MTBSO: estimation of the __________ of system outages. together states how long and frequency outages may be, but

availability estimate does not have that level of detail.

• More specific in how ______ system outages last and how ______ they occur.

• Example: 99.98% can be 15 minutes outage/day or 1.5 h outage/year

• General threshold: MTTR of 1 hour and MTBF/MTBSO of 4000 hours.

23

MTBF/MTBSO and Availability for MTTR of 1, 2, and 4 hours

Availability (% uptime)

MT

BF

/MT

BS

O (

Hou

rs)

MTTR4 hours2 hours1 hour

99.099.599.9

8000

4000

2000

24

Error and Loss Rates

• More complex than availability and recoverability• More commonly used the __________ level.

– Per _____ or circuit, such as BER– Between network equipment, such as CLR, CMR (cell

misinsertion ratio, between ATM equipments) or packet loss rates (between network layer routers).

– End-to-end, between hosts or applications.

• May be measured at the ____ and network layers, and reported as percentage of available traffic in the network.

25

Threshold for Reliability

• Good methods to estimate availability thresholds for your environment.– Evaluate the availability requirements of each of the

__________s that will be used in your environment, from discussions with users of the applications or from documentation for each application.

– Determine low-performance/high performance thresholds

– Estimate availability based on the probable _________ paths that the applications will use, and what equipment and services exist or may be in those paths.

26

Availability Threshold Summary

• Testbed or prototype reliability (availability): – less than ___

• Low-performance reliability: – less than ______%

• High-performance reliability: – greater than or equal to 99.9%

99.0 99.5 99.9 99.95

High-performance

Low-performancetestbed

27

Recoverability Threshold

• Low-performance reliability:– MTTR greater than __ hours or an

MTBF/MTBSO less than 8000 hours

• High-performance reliability:– MTTR less than or equal to __ hours and

MTBF/MTBSO greater than to equal to 8000 hours.

28

Error rates

• Low-performance– IP packet losses of

• >25% for < 2 hours/month

• 10% < packet loss < 25% for 2 hours/month

• 1% < packet loss < 10% for < 5 hours/month

• < 1 % for the remainder of the month

29

Delay Requirements

• Threshold for Delay– _________ delay (INTD)

• How long a user is willing to wait for a response from the system during an interactive session. (10 to 30 seconds)

• Important for interactive applications.

– Human _________ time (HRT)• The time boundary when users begin to perceive delay in the

system. (100ms)• Important for highly interactive applications.

– Network _________ delay• Lower bound to the end-to-end and round-trip network and

system delays.

30

Delay Estimates for User Requirements

Human Response Time

Network Propagation DelayInteraction Delay

0.01 0.1 1.0 10 100

31

Distinguishing between Burst and Bulk Applications

• Use ___ and ______ to distinguish between interactive-burst and interactive-bulk applications.

0.01 0.1 1.0 10 100

Interactive-Burst Interactive-Bulk/Burst Interactive-Bulk

Human Response Time Interaction Delay

Delay (Seconds)

32

Task Completion Time (TCT)Another approach for distinguishment

• The amount of time work is being performed by the system before __________ with the user is required.

Source Destination

Data received/processed

Data received/processed

Data received/processed

Delay

Delay

RTTDelay

TCT

Task completed

33

System Responsiveness

• System responsiveness = ________, when HRT/RTT > or =1 and

• System Responsiveness = HRT/__________, when HRT/RTT < ____– HRT/RTT describes the degree of responsiveness inherent

in the _______.– Small RTT, system responsiveness would be within the

______ time– Large RTT, delay would impact system responsiveness

• SR < __, interactive-bulk; SR >= 3, interactive _______.

34

Results of Delay Analysis

• Whether there are delay requirements for ___ of the applications, or whether all applications can use a ________ service.

• If there are delay requirement, how strict they are, in terms of:– ___________

– _________ levels (boundaries, thresholds, and guarantees)

distinguishing between applications that have real-time, interactive-burst, interactive bulk, and asynchronous delays.

35

End-to-End Delay

• End-to-end delay is useful in measuring and monitoring delay in an environment.

• It is based on combinations of – ___________ of the network

– Host and _______ element hardware and software performance, including network protocols

– Application behavior at particular delay thresholds

– User ________ with the system at particular delay threshold.

• Key guideline: determine the limiting factor between these delay thresholds (_________ within the system).

36

Delay Variation

• ______-arrival time of information

• A good rule of thumb is approximately 1% to ___% of the end-to-end delay.

37

Capacity Requirements

• Estimating data rate– Estimate ______ data rate (PDR), minimum data rate,

sustained data rate (SDR), or a combination of these.– Estimate a data size and ______s.– Different applications can be estimated differently based on

how much information you know about the ______________ characteristics (and data flow) of the application.

• _____ general thresholds between low- and high-performance capacity– It has a secondary impact on performance, through reliability

and delay.

38

Performance Envelop with Generic Thresholds

Low performanceregion

Delay(D)

Reliability (R)

Capacity (C )

High-PerformanceRegions

Generic ReliabilityThreshold

Generic Capacitythreshold

Generic DelayThreshold

39

Step 4: Developing Performance Thresholds

• Criteria for distinguishing between services– A general threshold– An environment-specific threshold – Specific services

40

Environment-specific Service Thresholds

• The ____________ indicates what the performance thresholds should be.

• Distinguish low- and high-performance characteristics for each environment (environment-specific thresholds), and not __________ characteristics across environment.

41

Comparing Application Characteristics

X ms Delay (ms)

App

lica

tion

s

High-performance Low performance

42

Sample Plot of a Characteristic as a Continuous Range of Values

X ms Delay (ms)

App

lica

tion

s

High-performance Low performance

Arbitrary Threshold

43

Step 4: Developing Performance Thresholds

• Criteria for distinguishing between services– A general threshold– An environment-specific threshold – Specific services

44

Specified Services

• Deterministic Services– More _________ performance characteristics

than the best-effort services.– Use _________ to approximate where the high

and/or low-performance levels are.– No guarantee is given

45

Guaranteed Services

• Need to have ___________ performance characteristics.

• There is some mechanism to ________ the service to the application or user.

• ________ is often useful to protect traffic flows downstream from a flow that exceeds its service limit and attempts to use more network resources than are contracted.

Time

Guarantee

Application behavior

Policing

No action taken

46

Fully Developed Performance EnvelopDelay(D)

Reliability (R)

Capacity (C ) Generic ReliabilityThreshold

Generic Capacitythreshold

Generic DelayThreshold

Service Boundary

Serviceguarantees

47

Distinguishing Between Service Performance Levels

• Guideline– Determine if any of the _________ have

obvious requirements for specified performance from the system.

– ______ application, such as mission-critical, real-time, or controlled rate.

– Apply application ______ to the applications.– Otherwise, most likely best effort.