Mälardalen University (MdH)

1

23-04-22

Component-based Approach and Dependable Systems

Concerning Predictability of Component-basedSystems: Classification of Quality Attributes

Ivica Crnkovic Mälardalen University, Sweden

Department of Computer Science and Engineeringwww.idt.mdh.se/~icc, [email protected]

2

23-04-22

Mälardalen University (MdH)Mälardalen University, Vasteras (Västerås)

Department of Computer Science and electronics (IDE)

Real-Time Systems Design Lab Computer Architecture Lab

Computer Science Lab Software Engineering Lab Intelligent sensors Lab

Medical Equipment

Prof. in Software Engineering http://www.idt.mdh.se/~icc [email protected]

3

23-04-22

4

23-04-22

Sources of information

Ivica Crnkovic and Magnus Larsson:

Building Reliable Component-Based Software Systems

Artech House Publishers, 2002, ISBN 1-58053-327-2

http://www.idt.mdh.se/cbse-book/

Ivica Crnkovic, Magnus Larsson Otto PreissConcerning Predictability in Dependable Component-Based Systems: Classification of Quality Attributes, Architecting Dependable Systems III,, p pp. 257 – 278, Springer, LNCS 3549, Editor(s): R. de Lemos et al. (Eds.):, 2005

ISO/IEC, “Software engineering - Product quality - Part1: Quality model”, ISO/IEC, International, Standard 9126-1:2001(E).

Ralf H. Reussner, Heinz W. Schmidt, Iman H. Poernomo, Reliability prediction for component-based software architectures The Journal of Systems and Software 66 (2003) 241–252

Claes Wohlin, Per Runeson: Certification of Software ComponentsIEEE Trans. Software Eng. 20(6): 494-499 (1994)

5

23-04-22

Outline

• Introduction– Why CBD

• Parti I– Properties (Quality attributes)

• Part II– Composability vs. predictability of Quality

Attributes– Classification of Quality attributes

6

23-04-22

What is Component-based approach?

• Building systems from (existing) software components– Providing support for the development of systems as assemblies

of components

– Supporting the development of components as reusable units

– Facilitating the maintenance and evolution of systems by customizing and replacing their components

– Separation of building components from building systems

7

23-04-22

Why component-based approach?

• Primary a concern of business and life-cycle factors– Costs, Time-to-market– Flexibility– Understandability, maintainability– Reuse of already existing software

• Higher abstraction level for functional properties• To less degree a concern of non-functional properties

– The requirements that must be fulfilled also with this approach– Sometimes more difficult to achieve– Might be a reason that component-based approach is less (or

not) feasible

8

23-04-22

The main question• CBD has shown to be successful in many domains

(desktop, distributed web-based applications...)

• What are the benefits and what are the drawbacks of CBD for particular domains?– Benefits /drawbacks for dependable (embedded) systems?

• What are the primary concerns and requirements of particular domains?– Can CBD contribute in providing solutions that meet these

requirements/concerns?• Yes• No• Make it more difficult

9

23-04-22

Part IComponents and system

propertiesWhat are properties?

10

23-04-22

Properties• Attribute/property

– “a construct whereby objects and individuals can be distinguished”

– “a quality or trait belonging to an individual or thing” • A required attribute/property is expressed as a need or

desire on an entity by some stakeholder. • An exhibited attribute/property is an attribute/property

ascribed to an entity as a result of evaluating (for example measurement of) the entity.

• The need for properties is motivated by their explanatory roles they have to fill. They describe phenomena of interest – There are no “absolute” properties

11

23-04-22

Some example of properties• Reusability, Configurability, Distributeability, Availability, Confidentiality,

Integrity, Maintainability, Reliability, Safety, Security, Affordability, Accessibility, Administrability, Understandability, Generality, Operability, Simplicity, Mobility, Nomadicity, Hardware independence Software, independence, Accuracy, Footprint, Responsiveness, Scalability, Schedulability, Timeliness, CPU utilization, Latency, Transaction, Throughput, Concurrency, Efficiency, Flexibility, Changeability, Evolvability, Extensibility, Modifiability, Tailorability, Upgradeability, Expandability, Consistency, Adaptability, Composability, Interoperability, Openness, Heterogenity, Integrability, Audibility, Completeness, , Conciseness, Correctness, Testability, Traceability, Coherence, Analyzability, Modularity, ….

Kazman, R., L. Bass, G. Abowd, M. Webb, “SAAM: A method for analyzing properties of software architectures,” Proceedings of the 16th International Conference on Software Engineering, 1994.

Kazman et al, Toward Deriving Software Architectures from Quality Attributes, Technical Report CMU/SEI-94-TR-10, 1994.

McCall J., Richards P., Walters G., Factors in Software Quality, Vols I,II,III', US Rome Air Development Center Reports, 1977.Bosch, J., P. Molin, “Software Architecture Design: Evaluation and Transformation,” Proceedings of the IEEE Conference and Workshop on Engineering of Computer-Based Systems, 1999.

12

23-04-22

Classification of properties• Different classification

– Run-time properties – Life cycle properties– Run time

• Reliability, safety, performance, robustness – Life cycle

• Maintainability, portability, reusability,…

• CBSE– Component properties– System properties

• Emerging properties

13

23-04-22

Quality model in ISO 9126-I

Examplehaving source code reviews” (a Software development process quality) influences the source code in that “the number of not initialized variables” (an internal quality attribute of a software product) is minimized. This positively influences the reliability, of the system (an external quality attribute of a software product).

14

23-04-22

General Concepts of the ISO/IEC 9126-1

Existing Components

16

23-04-22

ISO/IEC 9126-1 quality attributes

17

23-04-22

Other views – example: Dependability

1. Ability of a system to deliver service that can justifiably be trusted2. Ability of a system to avoid failures that are more frequent or more

severe than is acceptable to user(s)

Related to 1. Trustworthiness (assurance that a system will perform as expected)2. Survivability (capability to fulfill its mission in a timely manner)

Dependability

Safety-critical systems

Mission-critical systems

Business-critical systems

Avizienis, A.; Laprie, J.-C.; Randell, B.; Landwehr, C., “Basic concepts and taxonomy ofdependable and secure computing”, IEEE Trans. Dependable Sec. Comput., Vol. 1, Issue1, 2004

Other systems – embedded systems - Desktop systems

18

23-04-22

Dependability

Availability Reliability Safety Confidentiality Integrity Maintainability

Readinessfor usage

Continuity of services

Absence ofcatastrophic consequences

Absence ofunauthorized disclosure ofinformation

Absence of impropersystemalternations

Ability toUndergorepairs andevolutions

Attributes of Dependability

20

23-04-22

Dependability

Attributes

Threats

Means

AvailabilityReliabilitySafetyConfidentialityIntegrityMaintainability

Fault PreventionFault ToleranceFault RemovalFault Forecasting

FaultsErrorsFailures

21

23-04-22

Part II

Composition Predictability of properties

What do we know about properties compositions?

What do we need to know to predict system properties from component properties?

22

23-04-22

Main concern: Composability and predictability of extra-functional properties

• Classification in respect to:– to ability to predict system properties from

component properties• (ability to verify the predictability)

Ivica Crnkovic, Magnus Larsson, Otto PreissConcerning Predictability in Dependable Component-basedSystems: Classification of Quality AttributesArchitecting Dependable Systems II, Springer LNCS 2005

24

23-04-22

Some definitions first…

Assembly – a set of components

System

System Usage

System context

Component

25

23-04-22

Component development (COTS type)Known: Architectural Framework, component modelUnknown: system architecture, products, usage,..

Product lineKnown: domain, architectural framework, application skeleton,Variation (integration) pointsUnknown: Final products

Open systemsKnown: similar to PLA,but integrators are not necessary known

Final product ready to use(usage not necessary known)

Final product in use

What can we predict (or guarantee) about the system properties In each stage of development?

Different levels of knowledge about future component-based systems

26

23-04-22

Classification1. Directly composable properties. A property of an assembly

which is a function of, and only of the same property of the components involved.

2. Architecture-related properties. A property of an assembly which is a function of the same property of the components and of the software architecture.

3. Derived (emerging) properties. A property of an assembly which depends on several different properties of the components.

4. Usage-depended properties. A property of an assembly which is determined by its usage profile.

5. System context properties. A property which is determined by other properties and by the state of the system environment.

27

23-04-22

1. Definition: A directly composable property of an assembly is a function of, and only of the same property of the components.

• Consequence: to derive (predict) an assembly property it is not necessary to know anything about the system(s)

))(,),(),(()(}1:{

component assembly, attribute,

21 n

i

cPcPcPfAPnicA

cAP

28

23-04-22

Example

• “Physical characteristics”– Static memory

– (the “function” can be much more complicated)– (the functions are determined by different factors,

such as technologies)

componentsassembly,size,memory

)()(1

i

n

ii

cAM

cMAM

29

23-04-22

Example (cont)

• Dynamic memory – components with parameterized configurations/deployment paramentars

n

iicMAM

1

max )()(

30

23-04-22

2. Definition: An architecture-related property of an assembly is a function of the same property of the components and of the software architecture.

• Consequence: System/assembly architecture must be known

– Ok when building systems of particular class• (product-line architectures)

rearchitectu software)),(,),(),(()(

}1:{

21

SASAcPcPcPfAP

nicA

n

i

31

23-04-22

Clients

Client tier Web server tier Business logic tier Data tier

Web server

Businesscomponents

Data accesscomponents Data

Variabilitypoints

tionimplementaparticularaforfactorsalproportion,,componentsofnumber;clientsofnumber

ntransactiopertimeresponse/

/

cbayx

NT

cyyxbaxNT

Example - distributed systems

Yan L., Gorton I., Liu A., and Chen S., "Evaluating the scalability of enterprise javabeans technology", In Proceedings of 9th Asia-Pacific Software Engineer-ing Conference, IEEE, 2002.

32

23-04-22

3.Definition: A derived property of an assembly is a property that depends on several different properties of the components.

– Consequence: we must know different properties and their relations (might be quite complex) attributescomponent...

attributeassembly

)(,),(),(

),(,),(),(),(,),(),(

)(

}1:{

1

21

22212

12111

k

nkkk

n

n

i

PPP

cPcPcP

cPcPcPcPcPcP

fAP

nicA

33

23-04-22

C1wcet1f1

A

C2wcet2f2

Inputports

Outputports

end-to-end deadline is a function of different component properties, such as worst case execution time (WCET) and execution period.

Example

wcetcTccL

cBwcetccL jchpc j

in

iiin

ij

..

)()(.)(

)(

1

fixed priority scheduling

34

23-04-22

4. Definition: A Usage-dependent property of an assembly is a property which is determined by its usage profile.

Consequence: It is not enough to know which system will be built. It must be known how the system will be used

profileusagecomponent'profileusageassembly

profileusageparticularaforattribute

,:))'((),(

,

,,

ki

k

kiik

UUP

NkiUcPfUAP

35

23-04-22

Example Reliability• the probability that a system will perform its intended

function during a specified period of time under stated conditions.

• Mean time between failure• How to calculate reliability for Software System?

– Start from from a usage profile– Identify probability of the execution of components– Find out (measure) reliability of components– Calculate reliability of the system

Ralf H. Reussner, Heinz W. Schmidt, Iman H. Poernomo, Reliability prediction for component-based software architectures

The Journal of Systems and Software 66 (2003) 241–252

Claes Wohlin, Per Runeson: Certification of Software ComponentsIEEE Trans. Software Eng. 20(6): 494-499 (1994)

39

23-04-22

Uk

Ul

P(U)

Uk-min Uk-max

Ul -min Ul -max

Pl

Pk

Uk

Ul

P(U)

Uk-min Uk-max

Ul -min Ul -max

PlPl

PkPk

Can we predict reliability using existing usage profiles?Reuse problem:

mapping system usage profile to component usage profileWhen the known (measured) properties values can be reused?

),(),(),( maxmin kkllkkkl UAPUAPUAPUU

40

23-04-22

5. Definition: A System Environment Context property is a property which is determined by other properties and by context of the system environment.

– Consequence: It is not sufficient to know the systems and their usage, it is necessary to know particular systems and the context in which they are being performed

profile usageComponent ´System

contexttEnvironmen

profile; usage System

,,);),´((),(

,

,,,

ki

l

k

lkiiklkk

USE

U

NlkiEUcPfEUSP

41

23-04-22

Example

• safety property – related to the potential catastrophe– the same property may have different

degrees of safety even for the same usage profile.

42

23-04-22

Summary - Classification1. (DIR) - Directly composable properties. A property of an assembly which is a function of, and only of the

same property of the components involved.

2. (ART) - Architecture-related properties. A property of an assembly which is a function of the same property of the components and of the software architecture.

3. (EMG) - Derived (emerging) properties. A property of an assembly which depends on several different properties of the components.

4. (USG) - Usage-depended properties. A property of an assembly which is determined by its usage profile.

5. (SYS) - System context properties. A property which is determined by other properties and by the state of the system environment.

DIR – component context

DIR – Architecture (assembly) context

EMG – Architecture and other components context

USG – Use context

Sys – System (including external environemnt) context

44

23-04-22

Composition Combination of Different Types of Properties

• Classification should be– Complete– Orthogonal

• The shown classification is an idealization

– Some properties will be of combination of different types

45

23-04-22

46

23-04-22

Experiment

• Questionnaire– A set of defined properties– Grouped in different “sets of concerns”

• Following ISO 9126 standard and Dependability classification

• Assignment:– Classify the proposed properties

47

23-04-22

Questionnaire – instructions

For each property (Quality attribute) in the table ask the following questions:

a. (Directly composable attributes) Is it possible to analyze this assembly quality attribute given the same quality attribute of the components involved?

b. (Architecture Related attributes) Is it possible to analyze this assembly quality attribute given the assembly software architecture and the same quality attribute of the components involved?

c. (Derived attributes) Is it possible to analyze this assembly quality attribute from several different component attributes of the components involved?

d. (Usage-dependent attributes) Is it necessary to know the usage profile of the assembly to analyze this quality attribute?

e. (System environment context dependent attributes) Is it necessary to have system environment information to analyze this quality attribute?

Answer each question with:0 = No, 1 = yes, 2 = yes, significantly

There is also a question where you can indicate the confidence in your answers for each attribute. (scale 1 to 5)

Please fill in answers on all attributes and indicate with the confidence of your selection1= little confidence 5=high confidence

Questionnaire

48

23-04-22

Survey -Distribution of attribute classification

16%

44%13%

12%

15%a) Directly composable

b) Architecture-related

c) Derived

d) Usage-dependent

e) System environmentcontext

49

23-04-22

Level of agreement of the participants for the classification

57%

95%

66%74%

56%

0%

20%

40%

60%

80%

100%

Leve

l of a

gree

men

t

a) Directly composable

b) Architecture-related

c) Derived

d) Usage-dependent

e) System environmentcontext

51

23-04-22

Conclusion• “Return of investment” for component-based approach

depends also on predictability and assurance of quality attributes

• Composability of different quality attributes is different• The ability of predicting system properties from

component properties vary – from pure component properties to properties of systems exhibited differently in different context

• This limits the prediction ability of CB system’s behaviour