day 5 reliability

Upload: sbvseshagiri1407

Post on 28-Feb-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/25/2019 Day 5 Reliability

    1/77

    Software Testing and ReliabilityReliability and Risk Assessment

    Aditya P. Mathur

    Purdue UniversityAugust 121!

    " #uidant $or%orationMinnea%olis&St Paul' M(

    Graduate Assistants: Ramkumar NatarajanBaskar Sridharan

    Last update: August 16, 2002

  • 7/25/2019 Day 5 Reliability

    2/77

    Software Testing and 2

    Reliability and risk assessment )earning ob*e+tives

    1. ,hat is software reliability-

    2. ow to estimate software reliability-

    /. ,hat is risk assessment-

    0. ow to estimate risk using a%%li+ation ar+hite+ture-

  • 7/25/2019 Day 5 Reliability

    3/77

    Software Testing and 3

    Referen+es1. Statistical Methods in Software Engineering: Reliability and

    Risk' (oer . Sing%urwalla and Simon P. ,ilson' S%ringer'

    1333.

    2. Software Reliability, Measurement, Prediction,Application' 4ohn . Musa' Anthony 5annino' and

    6auhira 7kumoto' M+#rawill 8ook $om%any' 139:.

    /. A Methodology for Architecture Leel Reliability RiskAnalysis' S. M. ;a+oub and . . Ammar' 5

  • 7/25/2019 Day 5 Reliability

    4/77

    Software Testing and 4

    Software Reliability Softare re!ia"i!it# is the pro"a"i!it# of fai!ure free operation

    of an app!i$ation in a spe$ified operating en%ironmentand

    time period&

    Reliability is onequality metric. Others include

    performance, maintainability, portability, and

    interoperability

  • 7/25/2019 Day 5 Reliability

    5/77

    Software Testing and 5

    7%erating

  • 7/25/2019 Day 5 Reliability

    6/77

    Software Testing and 6

    Un+ertainty Uncertaintyis a common phenomena in our daily lives.

    In software engineering, uncertainty occurs in all phases of thesoftware life cycle.

    Examples:

    Will the schedule be met?

    What is the number of faults remaining faults?

    How many testers to deploy?

    How many months will it take to complete the design?

  • 7/25/2019 Day 5 Reliability

    7/77

    Software Testing and 7

    Probability and statisti+s Uncertainty can be quantified and managed using probability

    theory and statistical inference.

    Probability theory assists with quantification and combination of

    uncertainties.

    Statistical inference assists with revision of uncertainties in

    light of the available data.

  • 7/25/2019 Day 5 Reliability

    8/77

    Software Testing and 8

    Probability Theory In any software process there are known and unknown

    quantities.

    The knownquantities constitute history and is denoted by H.

    The unknown quantities are referred to as random

    quantities.

    Each unknown quantity is denoted by a capital letter such

    as TorX.

  • 7/25/2019 Day 5 Reliability

    9/77

    Software Testing and 9

    Random >ariables

    When a random quantity can assume numerical values it is

    known as a random variable.

    Specific values of TandXare denoted by lower case letters t

    andx and are known as realizationsof the corresponding

    random quantities.

    Example: IfXdenotes the outcome of a coin toss, thenXcan assume a value 0 (for head) and 1 (for tail).Xis a

    random variable under the assumption that on each toss the

    outcome is not known with certainty.

  • 7/25/2019 Day 5 Reliability

    10/77

    Software Testing and 10

    Probability

    P(E|H)

    or brevity we will su%%ressHand to denote the %robability

    of Eas sim%ly

    '(EP

    The %robability of an event E+om%uted at time in light of

    history H is given by

  • 7/25/2019 Day 5 Reliability

    11/77

    Software Testing and 11

    Random

  • 7/25/2019 Day 5 Reliability

    12/77

    Software Testing and 12

    8inary Random >ariables

    A discrete random variableis one whose realizations are

    countable.

    When e1and e2are numerical values, such as 0 and 1, thenEis

    known as a binary random variable.

    Example: Number of failures encountered over four hoursof application

    use.

    A continuous random variableis one whose realizations are not

    countable.

    Example: Time to next failure.

  • 7/25/2019 Day 5 Reliability

    13/77

    Software Testing and 13

    Probability distribution fun+tion

    '(xFX

    5f E is the event that'()x then is known as

    the distributionfun+tionof'and is denoted as

    '( xXP

    or a random variable'let Ebe the event that'Bx

    0'( >= xXP 5f then C is said to have a %oint mass.

    (ote that is nonde+reasing in D and ranges from = to

    1.'(xF

    X

  • 7/25/2019 Day 5 Reliability

    14/77

    Software Testing and 14

    Probability density fun+tion

    '(xFX'(xFX

    5f' is +ontinuous and takes all values in some interval I

    and is differentiable with res%e+t to*for all*in I'

    then is absolutely +ontinuous.

    '(xFX The derivative of at D is denoted by and is

    known as the %robability density fun+tion of'.

    '(xfX

    '(xfX dD is the a%%roDimate %robability that the random

    variable'takes on a value in the interval*and*+d*.

  • 7/25/2019 Day 5 Reliability

    15/77

    Software Testing and 15

    0.

    0

    ')( xf

    x

    xexXP

    = ')(

  • 7/25/2019 Day 5 Reliability

    16/77

    Software Testing and 16

    8inomial istribution

    Su%%ose that an a%%li+ation is eDe+uted ( times ea+h

    with a distin+t in%ut. ,e want to know the numberof

    in%uts''' on whi+h the a%%li+ation will fail.

    (ote that the %ro%ortion of the +orre+t out%uts is a measure of

    the reliabilityof the a%%li+ation.

    '+an assume values x B=' 1' 2'E'(. ,e are interested in the

    %robability that'Bx.

  • 7/25/2019 Day 5 Reliability

    17/77

    Software Testing and 17

    8inomial istribution G+ontd.H

    Nxppx

    NpxXP xNx ,&&&,0,'1('(')( ===

    Under +ertain assum%tions' the following %robability

    model' known as the 8inomial distribution' is used.

    ere % is the %robability that Ci B 1 for iB1'E'(. 5n other words' % is the %robability of failure of any single run.

    ''*(*+(*'( xNxNx

    N=

  • 7/25/2019 Day 5 Reliability

    18/77

    Software Testing and 18

    Poisson istribution

    &&&&&2,1,0,*')( ===

    xxexXP

    x

    ,hen the a%%li+ation under test is almost error free and is

    sub*e+ted to a large number of in%uts' then is large' I1pJ is

    small' and -.!p/is moderate. The above assum%tion leads to a sim%lifi+ation of the 8inomial

    distribution into the Poisson distribution given by the formula

  • 7/25/2019 Day 5 Reliability

    19/77

    Software Testing and 19

    Software Reliability@ Ty%es

    Re!ia"i!it# on a sing!e ee$ution: P(X-1)H', mode!ed "#

    Bernou!!i distri"ution&

    Re!ia"i!it# o%er N ee$utions: P(X-x)H', for -0,1,2,.N,

    gi%en "# Binomia! distri"ution or /oisson distri"ution for

    !arge Nand sma!! parameter %a!ue p&

    Re!ia"i!it# o%er an infinite num"er of ee$utions: P(X-x)H', for x-1,2,.N& Note that e are interested in the

    num"er of inputs after hi$h the first fai!ure o$$urs& his

    is gi%en "# geometri$ distri"ution&

  • 7/25/2019 Day 5 Reliability

    20/77

    Software Testing and 20

    Software Reliability@ Ty%es G+ontd.H

    hen the inputs to softare o$$ur $ontinuous!# o%er

    time, then e are interested in P(X-)H', i&e& the

    pro"a"i!it# that the first fai!ure o$$urs after time units&his is gi%en "# the eponentia! distri"ution&

    he time of o$$urren$e to the kth fai!ure $an "e gi%en "#

    the Gamma distri"ution&

    here are se%era! other mode!s of re!ia"i!it#, o%er one

    hundred*

  • 7/25/2019 Day 5 Reliability

    21/77

    Software Testing and 21

    Software failures@ Sour+es of un+ertainty

    Uncertainty about the presence and location of defects.

    Uncertainty about the use of run types. Will a run for

    a given input state cause a failure?

  • 7/25/2019 Day 5 Reliability

    22/77

    Software Testing and 22

    ailure Pro+ess

    5n%uts arrive at an a%%li+ation at random times.

    Some in%uts +ause failures and others do not.

    T1' T2' Edenote I$PUJ times between a%%li+ation

    failures.

    Most reliability models are +entered around the interfailuretimes.

  • 7/25/2019 Day 5 Reliability

    23/77

    Software Testing and 23

    ailure 5ntensity and Reliability

    ailure intensity is the number of failures eD%erien+ed

    within a unit of time. or eDam%le' the failure intensity

    of an a%%li+ation might be =./ failures&hr.

    ailure intensity is an alternate way of eD%ressing

    reliability' RIJ' whi+h is the %robability of no failures overtime duration .

    or a +onstant failure intensity we have RIJBe. 5t is safe to assume that during testing and debugging' the

    failure intensity de+reases with time and thus the reliability

    in+reases.

  • 7/25/2019 Day 5 Reliability

    24/77

    Software Testing and 24

    4elinski and Moranda Model G13:2H

    The a%%li+ation +ontains an unknown number of defe+ts.

  • 7/25/2019 Day 5 Reliability

    25/77

    Software Testing and 25

    4elinski and Moranda Model G+ontd.H

    Thus' given =BS=KBS1KBE.KBSi' iB1' 2E and some +onstant c' we obtain the following

    failure intensity' where S='S1'E'Si are su%%osed software failure times' failure rate rTiis

    given by@

    '1('( 1 += iNcStr iTi 1 iStfor

    Note that the failure rate

    drops by a constant amount.

    ttimeS0=0 S1 S2 S3

    '(tr

    http://../Macintosh%20HD/Users/apm/Desktop/im8zb816.ppt
  • 7/25/2019 Day 5 Reliability

    26/77

    Software Testing and 26

    Musa7kumoto Model@ Terminology

  • 7/25/2019 Day 5 Reliability

    27/77

    Software Testing and 27

    Musa7kumoto Model@ Terminology G+ontd.H

    (umber of inherent faults@ =BI I

    (umber of sour+e instru+tions@ I

    5nstru+tion eDe+ution rate@ r

  • 7/25/2019 Day 5 Reliability

    28/77

    Software Testing and 28

    Musa7kumoto@ 8asi+ Model

    ailure intensity for basi+ eDe+ution time model

    () = 0

    [1

    0 ]

    () =0e00

    R(' | ) = e{[0e

    0

    0

    ][1e0

    0 '

    ]}

  • 7/25/2019 Day 5 Reliability

    29/77

    Software Testing and 29

    Musa7kumoto@ )ogarithmi+ Poisson Model

    ailure intensity de+ay %arameter@

    ailure intensity for logarithmi+ Poisson model@

    () =0e

    () = 0

    0+1

    R(' |) =[

    0+1

    0('+)+1]1/

  • 7/25/2019 Day 5 Reliability

    30/77

    Software Testing and 30

    ailure intensity +om%arison as a fun+tion

    of average failures eD%erien+ed

    0

    0

    Average number of failures experienced

    F

    ailureintensity

    ()

    Logarithmic Poisson model

    Basic model

  • 7/25/2019 Day 5 Reliability

    31/77

    Software Testing and 31

    ailure intensity +om%arison as fun+tion

    of eDe+ution time

    0

    0

    Execution time

    Failureintensity

    ()

    Basic model

    Logarithmic Poisson model

  • 7/25/2019 Day 5 Reliability

    32/77

    Software Testing and 32

    ,hi+h Model to use-

    Uniform operational profile: Use the basic model

    Non-uniform operational profile: Use the logarithmic Poisson

    model

  • 7/25/2019 Day 5 Reliability

    33/77

    Software Testing and 33

    7ther issues

    Counting failures

    When is a defect repaired

    Impact of imperfect repair

  • 7/25/2019 Day 5 Reliability

    34/77

    Software Testing and 34

    5nde%endent +he+k against +ode +overage

    Reliability

    estimate

    Code coverage

    CL CH

    RL

    RH

    CL RL Unreliable estimate

    CL RH Unreliable estimate

    CH RL Reliable estimate

    CH RH Reliable estimate

  • 7/25/2019 Day 5 Reliability

    35/77

    Software Testing and 35

    7%erational Profile

    A quantitative characterization of how an application will be used. Thischaracterization requires a knowledge of input variables.

    Input stateis a vector of values of all input variables.

    Input variables:An interrupt is an input variable and so are all

    environment variables and variables whose values are input by the user

    via the keyboard or from a file in response to a prompt.

    Internal variables, computed from one or more input

    variables are not input variables.

    Intermediate results and interrupts generated during the execution as a

    result of the execution should not be considered as input variables.

  • 7/25/2019 Day 5 Reliability

    36/77

    Software Testing and 36

    7%erational Profile G+ontd.H

    Runs of an application that begin with identical input states belong to the same runtype.

    Example 1: Two withdrawals from the same person from the same

    account and of the same dollar amount.

    Example 2: Reservations made for two different people on the same

    flight belong to different run types.

    Function: Grouping of different run types. A function is conceived at

    the time of requirements analysis.

  • 7/25/2019 Day 5 Reliability

    37/77

    Software Testing and 37

    7%erational Profile G+ontd.H

    Function: A set of different run types. A function is conceived at the

    time of requirements analysis. A function is analogous to a use-case.

    Operation: A set of run types for the application that is built.

  • 7/25/2019 Day 5 Reliability

    38/77

    Software Testing and 38

    5n%ut S%a+e@ #ra%hi+al >iew

    Input space

    Inputstate

    Inputstate

    Inputstate

    Function 1

    Inputstate

    Inputstate

    Inputstate

    Function 2

    Function 3

    Function 4Function k

  • 7/25/2019 Day 5 Reliability

    39/77

    Software Testing and 39

    un+tional Profile

    un+tion Probability of o++urren+e

    1 =.!

    2 =./?

    / =.=?

  • 7/25/2019 Day 5 Reliability

    40/77

    Software Testing and 40

    7%erational Profile

    un+tion 7%eration Probability of o++urren+e

    1 711 =.0

    712 =.1

    71/ =.1

    2 721 =.=?

    722 =.1?

    / 7/1 =.1?

    7// =.=?

  • 7/25/2019 Day 5 Reliability

    41/77

    Software Testing and 41

    Modes and 7%erational Profile

    Mode un+tion 7%eration Probability of o++urren+e

    (ormal 1 711 =.0

    712 =.1

    71/ =.1

    2 721 =.=?

    722 =.1?

    / 7/1 =.1?

    7// =.=?

  • 7/25/2019 Day 5 Reliability

    42/77

    Software Testing and 42

    Modes and 7%erational Profile G+ontd.H

    Mode un+tion 7%eration Probability of o++urren+e

    Administrative A1 A711 =.0

    A712 =.1

    A2 A721 =.?

  • 7/25/2019 Day 5 Reliability

    43/77

    Software Testing and 43

    Reliability

  • 7/25/2019 Day 5 Reliability

    44/77

    Software Testing and 44

    Risk Assessment

    Risk is a $om"ination of to fa$tors: Probability of malfunction

    Consequence of malfunction

    3#nami$ $omp!eit# and $oup!ing metri$s $an "e used to a$$ount for thepro"a"i!it# of a fau!t manifesting itse!f into a fai!ure&

    Risk Assessment is usefu! in identif#ing: 5omp!e modu!es that need more attention

    /otentia! trou"!e spots

    7stimating test effort

  • 7/25/2019 Day 5 Reliability

    45/77

    Software Testing and 45

    Nuestion of interest

    Gi%en the ar$hite$ture of an app!i$ation, ho does one 8uantif# the riskasso$iated ith the gi%en ar$hite$ture

    Note that risk ana!#sis, as des$ri"ed here, $an "e

    performed prior to the de%e!opment of an# $ode and

    soon after the s#stem ar$hite$ture, in terms of its$omponents and $onne$tions, is a%ai!a"!e&

  • 7/25/2019 Day 5 Reliability

    46/77

    Software Testing and 46

    Risk Assessment Pro+edure

    3e%e!op S#stem Ar$hite$ture

    3etermine $omponent and $onne$tor $omp!eit#

    3e%e!op operationa! s$enarios and their !ike!ihood

    3e%e!op risk fa$tors

    /erform se%erit# ana!#sis

    /erform risk ana!#sis

    3e%e!op 53G

  • 7/25/2019 Day 5 Reliability

    47/77

    Software Testing and 47

    $ardia+ Pa+emaker@ 8ehavior Modes

    L1

    What is paced?

    A: Atrium

    V: Ventricle

    D: Dual; (both)

    Behavior mode indicated by 3-letter acronym: L1L

    2L

    3

    L2

    Which chamber is

    being monitored ?

    A: Atrium

    V: Ventricle

    D: Dual; (both)

    L3

    What is the mode type?

    I: Inhibited

    T: Triggered

    D: Dual pacing

    Example: VVI: Ventricle is paced when Ventricular sense does

    not occur, pace is Inhibited if a sense does occur

  • 7/25/2019 Day 5 Reliability

    48/77

    Software Testing and 48

    Pa+emaker@ $om%onents and $ommuni+ation

    Reed

    SwitchCommunication

    Gnome

    Coil

    DriverAtrial

    ModelVentricular

    Model

    enables

    enables

    heart

    magnet

    programming

  • 7/25/2019 Day 5 Reliability

    49/77

    Software Testing and 49

    $om%onent es+ri%tion

    Reed Swit+h IRSJ@Magneti+ally a+tivated swit+hL must be

    +losed before %rogramming +an begin.

    $oil river I$J@Pulsed to send =Os and 1Os by the %rogrammer.

    Atrial Model IARJ@$ontrols heart %a+ing.

    $ommuni+ations #nome I$#J@Re+eives +ommands as bytes

    from $ and send to AR and >T.

    >entri+ular Model I>TJ@$ontrols sensing and the refra+tory

    %eriod.

  • 7/25/2019 Day 5 Reliability

    50/77

    Software Testing and 50

    S+enarios

    Programming@ Programmer sets the o%eration mode of the devi+e.

    A>5@>T monitors the heart. ,hen a heart beat is not sensed the

    AR %a+es the heart and a refra+tory %eriod is in effe+t.

    AA5@ The AR +om%onent %a+es the heart when it does not sense any %ulse.

    >>5@>T +om%onent %a+es the heart when it does not sense any

    %ulse.

    >>T@>T +om%onent +ontinuously %a+es the heart.

    AAT@The AR +om%onent +ontinuously %a+es the heart.

  • 7/25/2019 Day 5 Reliability

    51/77

    Software Testing and 51

    Stati+ $om%leDity for 77 esigns

    $ou%ling 8etween $lasses I$8$J@Total number of other

    +lasses to whi+h a +lass is +ou%led.

    $ou%ling@ Two +lasses are +onsidered +ou%led if methods fromone +lass use methods or instan+e variables from other +lass.

  • 7/25/2019 Day 5 Reliability

    52/77

    Software Testing and 52

    7%erational $om%leDity for State+harts

    ynami+ +om%leDity fa+tor for ea+h +om%onent is based on +y+lomati++om%leDity of the state+hart s%e+ifi+ation for ea+h +om%onent.

    #iven a %rogram gra%h # with e edges and n nodes'

    the +y+lomati+ +om%leDity >I#JBenF2.

    or ea+h eDe+ution s+enario Ska subset of the state+hart s%e+ifi+ation ofthe +om%onent is eDe+uted thereby eDer+ising state entries' state eDits'

    and fired transitions.

    The +y+lomati+ +om%leDity of the eDe+uted %ath for ea+h +om%onent3i is +alled the operational comple*ity denoted bycp*k -3i/.

  • 7/25/2019 Day 5 Reliability

    53/77

    Software Testing and 53

    ealing with $om%osite States

    init t11

    s11

    Is1

    init

    s22

    Is2

    s21

    t13

    VGx (s11)

    +VGa (t11)

    +VGx (s1)

    +VGa (t12)

    +

    VGe(s1) +VG

    a(t13)+VGe(s22)

    Cyclomatic complexity for the s11 to s22 transition:

    VGp: p: x, a, e: Complexity of the exit, action, and entry code segments code segment

    t12

  • 7/25/2019 Day 5 Reliability

    54/77

    Software Testing and 54

    ynami+ $om%leDity for State+harts

    or ea+h eDe+ution s+enario these variables are u%dated with the

    +om%leDity measure of the thread that is triggered for that %arti+ulars+enario.

  • 7/25/2019 Day 5 Reliability

    55/77

    Software Testing and 55

    $om%onent $om%leDity Seuen+e diagrams are develo%ed fo ea+h s+enario.

  • 7/25/2019 Day 5 Reliability

    56/77

    Software Testing and 56

  • 7/25/2019 Day 5 Reliability

    57/77

    Software Testing and 57

    $onne+tor $om%leDity

    Simulation is used to determine the dynami+ +ou%ling measurefor ea+h +onne+tor.

    $ou%ling amongst +om%onents is re%resented in the form of amatriD.

    $ou%ling values are normalied to the highest +ou%ling.

  • 7/25/2019 Day 5 Reliability

    58/77

    Software Testing and 58

    $om%onent $om%leDity >alues

    RS $ $# AR >T

    Programming I=.=1J 9./ !:.0 20./

    A>5 I=.23J ?/.2 0!.9

    AAT I=.1?J 1==

    AA5 I=.2=J 1==

    >>5 I=.1?J 1==

    >>T I=.2=J 1==

    of ar+hite+ture+om%leDity

    .=9/ .!:0 .20/ ?=.209 09.?:2

    (ormalied .=.==2 =.=1/ =.==? 1 =.3!/

  • 7/25/2019 Day 5 Reliability

    59/77

    Software Testing and 59

    $ou%ling MatriD

    RS $ $# AR >T Prog. eart

    RS .==10 .==10

    $ .==/ .=11

    $# .==2 .==10 .==10

    AR .2? 1

    >T .2: .9:/

    Programmer .==10 .==!

    eart .12/ ./=:

  • 7/25/2019 Day 5 Reliability

    60/77

    Software Testing and 60

    Severity Analysis

    Risk fa+tors are asso+iated with ea+h +om%onent and +onne+tor

    by %erforming severity analysis. 8asi+ failure modeIsJ of ea+h +om%onent&+onne+tor and its effe+t on the

    overall system is studied using failure mode and effe+ts analysis IM

  • 7/25/2019 Day 5 Reliability

    61/77

    Software Testing and 61

    Severity Ranking

    $atastro%hi+ I=.3?J@ailure may +ause death or total system loss.

    $riti+al I=.:?J@ailure may +ause severe in*ury' %ro%ertydamage' system damage' or loss loss of %rodu+tion.

    Marginal I=.?J@ailure may +ause minor in*ury' %ro%erty damage' system damage'delay or loss of %rodu+tion.

    Minor I=.2?J@ ailure not serious enough to +ause in*ury' %ro%erty damage'or system damage but will result in uns+heduled maintenan+e or re%air.

    omain eD%erts assign severity indi+es Isvrty iJ to the severity

    +lasses.

  • 7/25/2019 Day 5 Reliability

    62/77

    Software Testing and 62

    euristi+ Risk a+tor

    The highest severity indeD IsvrtyiJ +orres%onding to a severity level offailure of a given +om%onent i' is assigned as its severity value.

    8y +om%aring the result of the simulation with the eD%e+tedo%eration' severity level for ea+h faulty +om%onent for a givens+enario is determined.

    A euristi+ Risk a+tor IhrfiJ is then +om%uted for ea+h

    +om%onent based on its +om%leDity and severity value.

    iii svrtycpxhrf =

  • 7/25/2019 Day 5 Reliability

    63/77

    Software Testing and 63

    M

  • 7/25/2019 Day 5 Reliability

    64/77

    Software Testing and 64

    MT Send in+orre+t+ommand Ie.g.To7ffinstead ofTo5dleJ

    5n+orre+tinter%retationof %rogrambytes

    5n+orre+to%eration modeand %a+ing of theheart. evi+e stillmonitored by the

    %hysi+ian'immediatemaintenan+ereuired.

    Marginal

  • 7/25/2019 Day 5 Reliability

    65/77

    Software Testing and 65

    $om%onent Risk fa+tors@ Using ynami+ $om%leDity

    RS $ $# AR >T

    ynami+ +om%leDity .==2 .=1/ .==? 1 .3!/

    Severity .2? .2? .? .3? .3?

    Risk fa+tor .===? .==/2? .==2? .3? .3109?

  • 7/25/2019 Day 5 Reliability

    66/77

    Software Testing and 66

    $onne+tor Risk fa+tors@ Using ynami+ $om%leDity

    RS $ $# AR >T Prog. eart

    RS .===/? .===/?

    $ .===:? .==2:?$# .===? .===: .===:

    AR .2375 .95

    >T .2565 .82935

    Prog. .===/? .==1?

    eart .11!9? .2916

  • 7/25/2019 Day 5 Reliability

    67/77

    Software Testing and 67

    $om%onent Risk fa+tors@ Using Stati+ $om%leDity

    RS $ $# AR >T

    $8$ =.0: =.9 1 =.! =.!

    Severity =.2? =.2? =.? =.3? =.3?

    Risk fa+tors based on$87

    =.113 =.2 =.? =.?: =.?:

  • 7/25/2019 Day 5 Reliability

    68/77

    Software Testing and 68

    $om%onent Risk fa+tors@ $om%arison

    Dynamic metrics better distinguish AR and VT components as highrisk when compared with RS, CD, and CG.

    Using static metrics, CG is considered at the same risk level as AR andVT.

    In pacemaker, AR and VT control the heart and hence are the highestrisk components which is confirmed when one computes the riskfactors using dynamic metrics.

  • 7/25/2019 Day 5 Reliability

    69/77

    Software Testing and 69

    $om%onent e%enden+y #ra%hs I$#sJ

    A CDG is described by setsNandEwhereN is a set of of nodes and E is a setof edges. sand tare designated as the start and termination nodes and belong toN.

    Each node n inN: , where Ci is the componentcorresponding ton,RCiis the reliability of Ci, andECi is the average

    execution time of Ci .

    Each edge einE: , where Tij is the transition from node

    CitoCj,RTij is the reliability of this transition, and PTi jis the transition

    probability.

    In the methodology described here, risk factorsreplace the reliabilitiesof components and transitions.

  • 7/25/2019 Day 5 Reliability

    70/77

    Software Testing and 70

    #eneration of $#s

    Estimate the execution probability of each scenario.

    For each scenario estimate the execution time of each component and

    then, using the probability of each scenario, compute the averageexecution time of each component.

    Estimate the transition probability of each transition.

    Estimate the complexity factor of each component.

    Estimate the complexity factor of each connector..

  • 7/25/2019 Day 5 Reliability

    71/77

    Software Testing and 71

    $# for the Pa+emaker Iall transition labels not shownJ

    t

    s

  • 7/25/2019 Day 5 Reliability

    72/77

    Software Testing and 72

    Reliability Risk Analysis

    Ar+hite+ture risk fa+tor is obtained by aggregating the risk fa+tors of individual+om%onents and +onne+tors.

  • 7/25/2019 Day 5 Reliability

    73/77

    Software Testing and 73

    Risk Analysis Algorithm7R %aths

    8readth eD%ansions +orres%ond to Q7R %aths. The risk fa+tors

    asso+iated with all nodes along the breadth eD%ansion aresummed u% weighted by the transition %robabilities.

    e1:

    e2:

    n1:

    n2:

    RB1GI1=.?J=./FI1=.!J=.:H.

    s

    n1 n2

    e1 e2

    Traverse the $# starting at node sand sto% until either tisrea+hed or the average a%%li+ation eDe+ution time is +onsumed.

  • 7/25/2019 Day 5 Reliability

    74/77

    Software Testing and 74

    Risk Analysis AlgorithmA( %aths

    The de%th of a %ath im%lies seuential eDe+ution. or eDam%le'su%%ose that node n.is rea+hed from node svia edge e.andthat node n4is rea+hed from n.via edgee4. Attributes of theedges and +om%onents are as follows@

    e1:

    e2:

    n1:

    n2:

    RB1GI1=.?J=./ D I1=.!J=.:H. TimeBTimeF? F12

    s

    n1

    n2

    e1

    e2

    The QA( %aths take into +onsideration the +onne+tor risk fa+tors Ihrfi*J

  • 7/25/2019 Day 5 Reliability

    75/77

    Software Testing and 75

    Pa+emaker Risk

    #iven the ar+hite+ture and the risk fa+tors asso+iated with+om%onents and +onne+tor' the risk fa+tor asso+iated with the%a+emaker is +om%uted to be a%%roD. =.3.

    This value of risk is +onsidered high. 5t im%lies that the%a+emaker ar+hite+ture is +riti+al and failures are likely to be+atastro%hi+.

    Risk analysis tells us that the >T and AR +om%onents are the highest risk +om%onents

    Risk analysis also tells us that the +onne+tors between >T' AR and heart +om%onents are the highest risk+om%onents

  • 7/25/2019 Day 5 Reliability

    76/77

    Software Testing and 76

    Advantages of Risk Analysis

    The $# is useful for the risk analysis of hierar+hi+al systems.Risks for subsystems +an be +om%uted. These +ould then beaggregated to +om%ute then risk of the entire system.

    The $# is useful for %erforming sensitivity analysis. 7ne +ouldstudy the im%a+t of +hanging the risk fa+tor of a +om%onent onthe risk asso+iated with the entire system.

    As the analysis is being done' most likely' %rior to +oding' one might+onsider revising the ar+hite+ture or use the same ar+hite+ture butallo+ate resour+es for +oding and testing based on individual riskfa+tors.

  • 7/25/2019 Day 5 Reliability

    77/77

    Summary

    Reliability' modeling un+ertainty' failure intensity' o%erational%rofile' reliability growth models' %arameter estimation.

    Risk assessment' ar+hite+ture' severity analysis' risk fa+tors'$#s.