applying the gnst to the engie it ral-final

103
Applying the Generalised Normalised Systems Theory to the Engie IT Reference Architecture Library Author: ir. Haerens Geert Promotor: Prof. dr. Jan Verelst Datum: May 15 th , 2016

Upload: geert-haerens

Post on 13-Feb-2017

32 views

Category:

Documents


1 download

TRANSCRIPT

  • Applying the Generalised

    Normalised Systems Theory to

    the Engie IT Reference

    Architecture Library

    Author: ir. Haerens Geert

    Promotor: Prof. dr. Jan Verelst

    Datum: May 15th, 2016

  • Applying the GNST to the Engie IT RAL

    ii | P a g e

  • Applying the GNST to the Engie IT RAL

    iii | P a g e

    Contents

    1 Executive Summary ................................................................................................................. 1

    2 Introduction ............................................................................................................................. 2

    2.1 Problem Statement .......................................................................................................... 2

    2.2 Research Question ........................................................................................................... 2

    2.3 Conceptual Model ............................................................................................................ 3

    2.4 Research Methodology .................................................................................................... 3

    3 IT Infrastructure and the Engie IT Reference Architecture Library ......................................... 7

    3.1 IT Infrastructure ............................................................................................................... 7

    3.1.1 Architecture Frameworks ......................................................................................... 7

    3.1.2 ArchiMate ................................................................................................................. 7

    3.1.3 OIAM ......................................................................................................................... 8

    3.1.4 GRAAL...................................................................................................................... 10

    3.2 Engie IT RAL .................................................................................................................... 11

    3.2.1 Concept ................................................................................................................... 11

    3.2.2 RAL as a Modular Structure .................................................................................... 13

    4 Artefact Requirements .......................................................................................................... 15

    5 Artefact Design ...................................................................................................................... 16

    5.1 General info on the Generalised Normalised Systems Theory ...................................... 16

    5.2 Contextualisation for Infrastructure Systems ................................................................ 18

    5.3 Artefact Description ....................................................................................................... 18

    5.3.1 Artefact input .......................................................................................................... 18

    5.3.2 Artefact Content ..................................................................................................... 18

    6 Demonstrating the Artefact .................................................................................................. 20

    6.1 Housing Infrastructure Basis Service .............................................................................. 20

    6.1.1 Manifestations of Concern, State, Version and Instance ....................................... 21

    6.1.2 Applying the Artefact to Housing 1.0...................................................................... 23

    6.1.3 Applying the Artefact to Housing 2.0...................................................................... 25

  • Applying the GNST to the Engie IT RAL

    iv | P a g e

    6.1.4 Applying the Artefact to Housing 3.0...................................................................... 27

    6.2 Hosting Infrastructure Basis Service .............................................................................. 29

    6.2.1 Manifestations of Concern, State, Version and Instance ....................................... 29

    6.2.2 Applying the Artefact to Hosting 1.0 ...................................................................... 31

    6.2.3 Applying the Artefact to Hosting 2.0 ...................................................................... 33

    6.3 Proxy Infrastructure Basis Service .................................................................................. 35

    6.3.1 Manifestations of Concern, State, Version and Instance ....................................... 35

    6.3.2 Applying the Artefact to Proxy ................................................................................ 37

    7 Evaluation of the Artefact qualitative analyses .................................................................. 39

    8 Evaluation of the Artefact feedback ................................................................................... 40

    8.1 Value of the Artefact ...................................................................................................... 40

    8.2 Observed CE ................................................................................................................... 40

    9 Artefact Value ........................................................................................................................ 42

    9.1 Value for RAL Engie IT .................................................................................................... 42

    9.2 Value for the Architect ................................................................................................... 42

    9.3 Value for the Generalized Normalised Systems Theory GNST ...................................... 43

    10 Conclusions and Reflections .................................................................................................. 44

    10.1 Conclusions..................................................................................................................... 44

    10.2 Reflections ...................................................................................................................... 45

    10.2.1 GNST knowledge as essential Architecture Skill ..................................................... 45

    10.2.2 Artefact evolution ................................................................................................... 45

    10.2.3 Roadmaps and Lehmans Law ................................................................................. 45

    10.2.4 Configuration is the new Code ............................................................................... 46

    Appendix 1: Applying the GNST on the RAL/ASC Engie IT ............................................................ 47

    A1.1 Module: Housing-IBS......................................................................................................... 47

    A1.1.1 Housing 1.0 ................................................................................................................. 49

    A1.1.2 Module: Housing 2.0 .................................................................................................. 55

    A1.1.3 Module: Housing 3.0 .................................................................................................. 59

    A1.1.4 Conclusion and reflections on Housing-IBS................................................................ 64

  • Applying the GNST to the Engie IT RAL

    v | P a g e

    A1.2 Module: Hosting-IBS ......................................................................................................... 66

    A1.2.1 Hosting 1.0.................................................................................................................. 68

    A1.2.2. Hosting 2.0 ................................................................................................................ 70

    A1.2.3 Conclusions and reflections on Hosting pattern ........................................................ 73

    A1.3 Module: Proxy-Network-IBS ............................................................................................. 75

    A1.3.1 Cohesion ..................................................................................................................... 76

    A1.3.2 Coupling ...................................................................................................................... 76

    A1.3.3 Separation of Concern (SoC) ...................................................................................... 77

    A1.3.4 Separation of State (SoS) ............................................................................................ 77

    A1.3.5 Version Transparency (VT) ......................................................................................... 77

    A1.3.6 Instance Traceability (IT) ............................................................................................ 77

    A1.3.7 Adding a Source and Target ....................................................................................... 77

    A1.3.8 Conclusions ................................................................................................................. 78

    Appendix 2: CE in other IT Infrastructure Systems ....................................................................... 79

    A2.1 Housing ............................................................................................................................. 79

    A2.2 Upgrades OS - Hosting ...................................................................................................... 79

    A2.3 New workstation ............................................................................................................... 80

    A2.4 Unified Communication Tool ............................................................................................ 80

    A2.5 Infrastructure Software Update - Exchange 2013 ............................................................ 81

    A2.6 Instance Transparency ...................................................................................................... 81

    A2.7 AD Upgrade ....................................................................................................................... 81

    A2.8 Application functionality ................................................................................................... 81

    A2.9 Separation of State ........................................................................................................... 82

    Appendix 3: Infrastructure Solutions for Evolvability ................................................................... 83

    A3.1 Updates ............................................................................................................................. 83

    A3.2 Version Transparency ....................................................................................................... 84

    A3.3 Separation of State ........................................................................................................... 85

    References .................................................................................................................................... 87

    Glossary ......................................................................................................................................... 89

  • Applying the GNST to the Engie IT RAL

    vi | P a g e

    List of Figures

    Figure 1: Research conceptual model ............................................................................................. 3

    Figure 2: Design Science Process .................................................................................................... 4

    Figure 3: Archimate Technology Layer Meta Model ...................................................................... 8

    Figure 4: TOGAF Enterprise Continuum .......................................................................................... 9

    Figure 5: Classification of architecture disciplines according to NGI............................................ 10

    Figure 6: ASC part of RAL .............................................................................................................. 12

    Figure 7: Dependencies in the ASC part of the RAL ...................................................................... 14

    Figure 8: Artefact Requirements .................................................................................................. 15

    Figure 9: Generalization of the Normalised Systems Theory - Peter De Bruyn ........................... 17

    Figure 10: Housing 1.0 Modular Structure ................................................................................... 23

    Figure 11: Effect of change to Housing 1.0 ................................................................................... 24

    Figure 12: Housing 2.0 Modular Structure ................................................................................... 25

    Figure 13: Power Distribution Units, Patch panels and cable trays ............................................. 26

    Figure 14: Housing 3.0 Modular Structure ................................................................................... 27

    Figure 15: Microsoft seaborne data centre container .................................................................. 28

    Figure 16: Hosting 1.0 modular Structure .................................................................................... 31

    Figure 17: Hosting 2.0 modular Structure .................................................................................... 33

    Figure 18: Proxy modular Structure .............................................................................................. 37

    Figure 19: Housing 1.0 modular Structure .................................................................................... 49

    Figure 20: going from BNC to CAT3/5/6 cables ............................................................................ 51

    Figure 21: Different SCSI cables and connectors .......................................................................... 52

    Figure 22: from SCSI to FC and from From SC to LC Fiber connectors ......................................... 52

    Figure 23: Result unstructured cabling ......................................................................................... 54

    Figure 24: Housing 2.0 modular Structure .................................................................................... 55

    Figure 25: Patch panels and power distribution units .................................................................. 56

    Figure 26: Housing 3.0 modular structure .................................................................................... 59

    Figure 27: Zone concept ............................................................................................................... 59

    Figure 28: Hot and cold air distribution in Zone ........................................................................... 60

    Figure 29: Cooling of a zone .......................................................................................................... 60

    Figure 30: Top of Rack concept, including redundancy ................................................................ 61

    Figure 31: Data Center of the future according to Microsoft ....................................................... 65

    Figure 32: Hosting 1.0 modular structure ..................................................................................... 68

    Figure 33: Hosting 2.0 modular structure ..................................................................................... 71

    Figure 34: Proxy modular structure .............................................................................................. 76

  • Applying the GNST to the Engie IT RAL

    vii | P a g e

    List of Tables

    Table 1: OIAM Building blocks ........................................................................................................ 9

    Table 2: ASC as part of RAL ........................................................................................................... 12

    Table 3: Normalization Principles ................................................................................................. 17

    Table 4: Manifestations of C, S, V and I ........................................................................................ 18

    Table 5: GNST Compliancy check summary .................................................................................. 19

    Table 6: Housing 1.0 GNST Compliance summary ........................................................................ 24

    Table 7: Housing 2.0 GNST Compliance summary ........................................................................ 26

    Table 8: Housing 3.0 GNST Compliance summary ........................................................................ 28

    Table 9: Hosting 1.0 GNST Compliance summary......................................................................... 32

    Table 10: Hosting 2.0 GNST Compliance summary ...................................................................... 34

    Table 11: Proxy GNST Compliance summary ................................................................................ 38

    Table 12: Expert Team Evaluation 11 of the 13 Expert Team members ................................... 39

    Table 13: CE at different layers of the RAL ................................................................................... 41

    Table 14: Ethernet speed evolution.............................................................................................. 63

    Table 15: Max cable length for Ethernet ...................................................................................... 63

  • Applying the GNST to the Engie IT RAL

    viii | P a g e

  • Applying the GNST to the Engie IT RAL

    ix | P a g e

    Acknowledgements

    Glorie Deze reis omvatte veel emoties Midden in de grote ommezwaai van mijn leven Doorzetten met veel devotie Het heeft mij pure zuurstof gegeven Vele mensen moet ik mijn dank betuigen Zonder hen was het niet gelukt Lofzangen zal ik uit mijn duim zuigen Onder het wierookvat ga ik gebukt Wie beter om mee te starten Dan hij die weet wat mijn schrijven behelst Graag zou ik met hem normalized biljarten Mijn oprechte dank aan Jan Verelst Engie moet ik ook vermelden Onder de vorm van een manager, nen echte struise Begrepen wat ik hem vertelde deed hij zelden, Maar toch, merci Paul Buyse Zoveel toffe mede studenten Ik zal hen 1 voor 1 missen Harten van koekebrood met krenten Uit mijn geheugen zijn ze niet te wissen Het professoren korps van AMS Ben ik dankbaar voor het delen van hun kennis Van een aantal kreeg ik lichte stress Oeps, dit is heiligschennis Namaan, Frdric and Jenny I salute you for your generous feedback Your input was worth more than a penny My French normalized infrastructure stack

  • Applying the GNST to the Engie IT RAL

    x | P a g e

    Philippe en Christel, Jullie woorden ware pure erkenning Fijn dat jullie tijd stopten in mijn epistel Sentimenten gespeend van gewenning

    Een dankje voor de Koen Motor van de UCC infrastructuur Waar is de kok van toen Hij kookt nu IT met passie en vuur Christof ontmoette nimmer Frank Maar toch waren zij mijn NST referenties Van hen hoorde ik een andere klank Of ware het de foute moppen over haarextensies Marc en Dirk, de nestors Hun feedback was niet om mee te lullen Met hun ideen uiteengezet als volleerde lectors Ik kon er nog 2 thesissen mee vullen De wetenschap bespreek ik met Frederik, Een wijzer man is er niet te vinden Mijn besognes deel ik met Cdric, Ja die 2 zijn echte vrienden Zonder Stefan was het iets anders geworden Voor de verandering gaf hij mij structuur In hart en rede zijn wij verbonden Dankbaarheid tot voorbij ons laatste uur Waar was ik geweest zonder moeder aan mijn zij Terwijl ik de student was uit gaan hangen Droeg zij zorgt voor die 2 parels van mij Voor jouw zijn al mijn koor- en lofzangen En nu mijn liefste meisjes Richt pappie een woordje tot jullie Kleurders van mijn leven met zomer ijsjes De zin, mijn zijn, mijn ware glorie

    Drs G

  • Applying the GNST to the Engie IT RAL

    1 | P a g e

    1 Executive Summary

    Businesses are required to be more and more agile to compete in todays marketplace. An agile

    Business requires agile IT. Agile IT is an IT which is able to change quickly and without

    introducing a massive amount of unwanted side effects, as those will undermine the agility.

    The Generalized Normalized Systems Theory (GNST) studies the effect of change on modular

    structures and has come up with 4 principles which, when followed, allow the creation of

    modular structures which will stay stable no unwanted side effect know as Combinatorial

    Effects (CE) - under change.

    IT is not only about applications and software, there is also IT Infrastructure, the backbone on

    which all applications run. How well can IT Infrastructure handle change?

    This thesis is about applying the General Normalized Systems Theory (GNST) on IT

    Infrastructure. When IT Infrastructure is represented by a modular system, the GNST principles

    can be tested. A tool is being created to test and evaluate the compliance of a modular system

    representing an IT Infrastructure System, with the GNST. The Engie IT Reference Architecture

    Library (RAL) is used as a source of definitions of IT Infrastructure Systems.

    When an IT Infrastructure System is not compliant with the 4 GNST principles, Combinatorial

    Effects (CE) - hidden coupling or dependencies in a system which increase with the size of the

    system - are to be expected when the IT Infrastructure System changes.

    The thesis shows the GNST can be applied for IT Infrastructure Systems represented by a

    modular structure. The non-compliance with the GNST predicts CE which are practically

    observed. Multiple examples of CE are given and explained by means of the GNST principles.

    Besides the demonstration that the GNST can be applied to IT Infrastructure Systems, the thesis

    also demonstrates the usage of the GNST in a new field IT Infrastructure.

    The value of the thesis lies in recognizing that knowing and applying the GNST on IT

    Infrastructure Systems is an essential skill for an Architect as it allows him/her to create

    scientifically proven roadmaps taking into account the impact of future evolutions and change

    on the IT System.

  • Applying the GNST to the Engie IT RAL

    2 | P a g e

    2 Introduction

    2.1 Problem Statement

    Today it is all about Agility. Businesses need to react fast on changing conditions of the

    marketplace and on what the competition is doing. Businesses change their offerings, services,

    restructure internally to be better armed for these turbulent times.

    Agility is required at all levels and domains of the organization. The only constant is change. All

    this has impact on the IT landscape: Applications as well as Infrastructure. A lot of literature can

    be found on creating and changing applications in an agile way, but little guidance, exists with

    regards to IT Infrastructure, except the trendy statement put it in the cloud

    Can extra guidance be given at the IT Infrastructure level on how to create systems which can

    cope with change or can the impact of change be investigated during design time and/or

    runtime?

    The Generalised Normalised Systems Theory (GNST) is a theory about the behaviour of Modular

    Structures under change. Can the Generalised Normalised Systems Theory (GNST) be used to

    provide guidance on the design of IT Infrastructure Systems under change?

    2.2 Research Question

    Can the Generalised Normalised Systems Theory (GNST) be used to study the behaviour of IT

    Infrastructures under change?

    This leads to the following research questions:

    Part 1: Can the Generalised Normalised Systems Theory (GNST) principles be applied to

    modular structures representing IT Infrastructure?

    Part 2: Can Combinatorial Effects (CE) - hidden coupling or dependencies in a system which

    increase with the size of the system - be predicted and recognized at the IT Infrastructure level

    by using the Generalised Normalised Systems Theory (GNST)?

    Part 3: What is the added value of using the Generalised Normalised Systems Theory (GNST)

    at the IT Infrastructure level?

  • Applying the GNST to the Engie IT RAL

    3 | P a g e

    2.3 Conceptual Model

    Figure 1: Research conceptual model

    The Engie IT Reference Architecture Library (RAL, see section 3.2) contains an extensive list of

    Infrastructure Systems. Of some of those Infrastructure Systems, a detailed modular structure

    will be created. Based on this modular structure the Generalized Normalized Systems Theory

    (GNST) will be applied with special attention to compliance with the 4 GNST Principles (see

    section 5.1). If the modular structure is not compliant with the 4 GNST Principles, CE

    Combinatorial Effects are to be expected.

    Based on knowledge of the IT Infrastructure Reality of the studied Infrastructure Systems, CE

    can be observed in real life. Those CE can be linked to the predicted CE.

    By doing so, the applicability of the GNST for IT Infrastructure can be confirmed.

    2.4 Research Methodology

    The research will apply the GNST principles to IT Infrastructure. Design Science has been chosen

    as approach to conduct this research. Paul Johannesson and Erik Perjons1 propose a process to

    perform Design Science. The result of this Design Science process is an Evaluated Artefact.

    1 An Introduction to Design Science Paul Johannesson, Erik Perjons - 2014

  • Applying the GNST to the Engie IT RAL

    4 | P a g e

    Figure 2: Design Science Process

    The Artefact is a Generalised Normalised Systems Theory (GNST) analysis tool, demonstrated

    on IT Infrastructure modular structures and evaluated by an Expert Team.

    Explicate Problem is handled in Chapter 3 and will focus on the context of the problem

    statement, will discuss what IT Infrastructure is and what the Engie IT Reference Architecture

    Library (RAL) is about.

    Define Requirements is handled in Chapter 4 and will focus on what is expected from the

    Artefact to address the Research Question.

    Design and Develop Artefact is handled in Chapter 5 and focus on the creation of an Artefact

    which can be used to apply the Generalised Normalised System Theory (GNST) at the

    Infrastructure layer, and thus providing an answer to Part 1 of the Research Question (Can the

    GNST principles be applied to modular structures representing IT Infrastructure ?)

  • Applying the GNST to the Engie IT RAL

    5 | P a g e

    Demonstrate Artefact is handled in Chapter 0 and will focus on applying the Artefact to three

    Infrastructure modular structures and thus provide an answer to Part 2 of the Research

    Question (Can CE be predicted and recognized at the IT Infrastructure level by using the GNST ?)

    Evaluate Artefact is handled in Chapters 7, 8 and 9

    Chapter 7 will focus on a qualitative analyses of the Artefact demonstration by an Expert team,

    and on the quality of the answer to Part 2 of the Research Question.

    Chapter 8 will focus on additional feedback by the Expert team with regards to the Artefact and

    its applicability in IT Infrastructure

    Chapter 9 will focus on the value the Artefact has to IT Infrastructure architecture and to the

    Reference Architecture Library (RAL) of Engie IT and thus providing an answer to Part 3 of the

    Research Question (What is the added value of using the GNST at the IT Infrastructure level ?)

    Chapter 10 will contain the overall conclusion of the Research and additional reflections.

    Research Strategies and Methods, Creative Methods are based on the following:

    The Generalised Normalised Systems Theory (GNST)

    The Engie IT Reference Architecture Library (RAL)

    IT Infrastructure knowledge of the Expert Team

    ArchiMate and Open Infrastructure Architecture Method (OIAM)

    Accumulated knowledge of the writer

    Knowledge Base

    The GNST has never been applied to an IT Infrastructure modular structures. As such, there is

    no formal knowledge base against which the Artefact can be evaluated.

    In this research the Artefact will be evaluated by an Expert Team. The accumulated knowledge

    in the field of IT Infrastructure of this team will serve as Knowledge Base.

    The Expert team is asked to evaluate:

    The Correctness of the IT Infrastructure Modular Structure that is being investigated.

    The Correctness of the Analyses done in the Artefact (applying the GNST principles).

    Value of the Artefact.

    The Expert Team consists of people who have:

    IT Infrastructure expertise for at least 10 years

    Knowhow of the GNST and NST for at least 4 years

    IT Infrastructure Management expertise for at least 10 years

  • Applying the GNST to the Engie IT RAL

    6 | P a g e

    Expert Team:

    Frederik Leemans: IT Enterprise Architect Alkasa consulting

    Dirk Timmerman: Infrastructure Architect - Athos

    Cdric Carrette: Infrastructure Architect - Carrette GCV

    Koen Van den Broeck: Infrastructure Architect - Engie IT

    Marc Kimpe: Infrastructure Architect - Cronos

    Frank Van der Veken: Security Architect - Engie BeNeLux

    Christophe De Clercq: Application Architect - Contribute group (Cronos)

    Stefan Thys: Project Manager - Quantum Management and Consulting

    Christel Plessers: CIO - Mercedes Benz France

    Philippe Le Cerf: CIO - Vinotte

    Frdric Deleage: Infrastructure Architect Engie IT

    Jenny Edouard: Infrastructure Architect Engie IT

    Namaan Boutighane: Infrastructure Architect Engie IT

  • Applying the GNST to the Engie IT RAL

    7 | P a g e

    3 IT Infrastructure and the Engie IT

    Reference Architecture Library

    Before the Artefact can be applied at the level of IT Infrastructure, there must be a common

    understanding of what IT Infrastructure is. As the Artefact will be demonstrated on IT

    Infrastructure modular structures defined in the Engie IT Reference Architecture Library (RAL),

    the meaning and content the Engie IT RAL will be explained.

    3.1 IT Infrastructure

    The next sections will work towards a definition of IT Infrastructure.

    3.1.1 Architecture Frameworks

    IT Infrastructure or Infrastructure for short has less focus in Architecture Frameworks and

    Methodologies compared to the Application and Business components. Commonly used

    Architecture frameworks such as TOGAF do not go into details when it comes down to

    Infrastructure.

    TOGAF2 refers to the Technical layer when it comes down to Infrastructure and is not alone in

    this.

    The British Computer Society's "Reference Model for Enterprise and Solution Architecture"3

    defines different Types of Architecture and sees IT Infrastructure Architecture as follows:

    Technical architecture or infrastructure architecture: The structure and behaviour of the technology

    infrastructure. Covers the client and server nodes of the hardware configuration, the infrastructure

    applications that run on them, the infrastructure services they offer to applications, the protocols and

    networks that connect applications and nodes.

    This shifts the question from What is Infrastructure? to What is Technical?

    3.1.2 ArchiMate

    Like TOGAF, ArchiMate4 talks about the Technology Layer in its framework and about

    Infrastructure components in its Meta Model.

    This Meta Model contains abstract constructs which can represent infrastructure components.

    2 TOGAF Version 9 The Open Group

    3 en.wikipedia.org/wiki/Architecture domain

    4 ArchiMate 2.0 The Open Group

    https://en.wikipedia.org/wiki/Technical_architecture

  • Applying the GNST to the Engie IT RAL

    8 | P a g e

    Figure 3: ArchiMate Technology Layer Meta Model

    The Technology Layer of ArchiMate describes the Infrastructure in a formalized language.

    What ArchiMate, does not provide, is a proper definition on what Infrastructure actually is.

    3.1.3 OIAM

    The Open Infrastructure Architecture Method (OIAM)5, uses the concept of Infrastructure

    Functions of ArchiMate to build a generic library of Infrastructure functions. OIAM has built this

    library based on practical experience and the library is maintained by a user community. With

    the Infrastructure Functions of the Library, Infrastructure Patterns can be made, which are an

    aggregation of the Infrastructure Functions.

    OIAM is structured like the Enterprise Continuum of TOGAF with Infrastructure Functions as the

    generic building blocks part of the Architecture Continuum:

    The Patterns are the generic solutions of the Solution Continuum,

    The Specific Patterns as specific solutions of the Solution Continuum,

    Deployed Patterns as part of the Deployed Solutions Continuum.

    5 OIAM - www.infra-repository.org

  • Applying the GNST to the Engie IT RAL

    9 | P a g e

    Figure 4: TOGAF Enterprise Continuum

    List of Infrastructure Functions according to OIAM

    Table 1: OIAM Building blocks

    OIAM is one of the few generally available Infrastructure Framework which provides:

    A library of Infrastructure primary concepts (the functions),

    A modelling method by using ArchiMate,

    A way to focus on Infrastructure Function instead of Infrastructure Construction,

    A way to treat Infrastructure from a functional point of view as often done at the Business and

    Application layers.

  • Applying the GNST to the Engie IT RAL

    10 | P a g e

    3.1.4 GRAAL

    In the book Competences of the IT Architect, Wieringa et al6 try to summarize the key

    competences of different IT Architecture types, but quickly conclude that, as different sources

    use different definitions for both the Architecture domains and Architect classifications, they

    require a reference framework to structure their study. The GRAAL7 framework is used to

    position the different Architecture domains. With the domains being fixed they compare

    different Architect classifications or profiles to see which profile covers which part of the

    domains.

    The GRAAL framework uses the following layering:

    Business environment (BE): entities in the environment of the organisation to which the

    organisation delivers products and/or services. For commercial companies, the most important

    element types of the business environment are their customers.

    Business (B): the products and services that the organisation produces for its environment, the

    processes that create these products and services, the employees who perform those processes,

    the formal and informal relations between those employees, etc.

    Enterprise software systems (ESS): organisation specific software systems that support the

    processes and people in the business.

    Software infrastructure (SI): software systems that are not specific for the organisation, such as

    operating systems, database management systems, email servers, etc.

    Physical infrastructure (PI): processors, disks, network routers, switches and cables, and all other

    physical objects that are needed to run the software systems that constitute the business

    systems and software infrastructure layers.

    Figure 5: Classification of architecture disciplines according to NGI

    The NGI (Nederlands Genootschap voor Informatica) has plotted on this framework their

    classification of architecture disciplines/profiles. By doing so a clear definition of Infrastructure

    Architecture and the profile of Infrastructure Architect becomes visible.

    6 Competences of IT Architects Roel Wieringa, Pascal van Eck, Claudia Steghuis, Erik Proper - 2009

    7 graal.ewi.utwente.nl

  • Applying the GNST to the Engie IT RAL

    11 | P a g e

    The combination of GRAAL and NGI bring a definition of Infrastructure which is exactly how it is

    treated within Engie IT:

    Infrastructure = Physical Infrastructure (according to GRAAL) + Software Infrastructure (according to GRAAL)

    Hence, Infrastructure Architecture is about the structure of an Infrastructure system, its

    components, their relationship and principles guiding their evolutions8. Infrastructure

    Architecture is a conscious restriction of Infrastructure design space9.

    3.2 Engie IT RAL

    3.2.1 Concept

    Since 2009 the group of Infrastructure Architects at Engie IT has been using ArchiMate as

    modelling language for all Infrastructure models.

    The Infrastructure Usage View (IUV) is a view predefined in ArchiMate which focuses on how

    Infrastructure services are realised and how they are used by Applications. This view is very

    relevant as Engie IT delivers Infrastructure Services to the internal Business Units of Engie.

    Although a good starting point in finding a unified way of working and modelling amongst

    Infrastructure Architects, it became clear that models and the names given to the modelling

    concepts largely remained un-standardised. Additional normalization resulted in 2011 in the

    introduction of the RAL. The RAL is the Reference Infrastructure Library and includes all

    conceptual and deployed Architectures. The RAL is a simplified implementation of the

    Enterprise Continuum of TOGAF applied at the Infrastructure layer.

    Where the starting point of OIAM is Infrastructure Functions, the starting point of the RAL is

    Infrastructure Services. At the time the RAL was created, ArchiMate 2.0 did not exist yet and

    neither did the concept of the Infrastructure Function. The idea behind the RAL and OIAM is

    identical: a library of Architecture Building Blocks which can be combined to create new

    Architecture Patterns - this library is referred to as the Architecture Solutions Continuum (ASC).

    When those Building Blocks and Patterns are being deployed, they become part of the

    Deployed Solutions Continuum (DSC).

    The Architecture Solutions Continuum (ASC) contains the conceptual Infrastructure services and

    how they should be build,

    The Deployed Solutions Continuum (DSC) contains the deployment of such services in a specific

    context (for one of the BUs Engie IT creates services for).

    8 Defining Architecture IEEE 1471

    9 Enterprise Governance and Enterprise Engineering Jan A.P. Hoogervorst - 2009

  • Applying the GNST to the Engie IT RAL

    12 | P a g e

    The Architecture Solution Continuum can be split into 3 large categories:

    IBS Infrastructure Basic Services

    Contains the most elementary Infrastructure Services required to run an application and manage a data centre.

    Examples: Housing, Hosting, Network, Tooling, Monitoring, etc.

    IPS Infrastructure Platform Services

    Contains the Infrastructure Services which are the aggregation of Infrastructure Basic Services and software on which applications can be deployed.

    Examples: DB platform, WAS platform, SAP platform, etc.

    ISS Infrastructure Solution Services

    Contains the Infrastructure Services which are the aggregation of Infrastructure Basic Services, Platform Services and Software.

    Examples: Remote Access, Mail, Unified Communication, Voice over IP.

    Table 2: ASC as part of RAL

    Figure 6: ASC part of RAL

    Note that this corresponds entirely with what GRAAL defines as the Physical and Software

    Infrastructure:

    IBS 1. Physical Infrastructure and Software Infrastructure

    IPS 2. A combination of Physical and Software Infrastructure on which

  • Applying the GNST to the Engie IT RAL

    13 | P a g e

    Enterprise Software Systems can be deployed

    ISS 3. Software Infrastructure

    A specific naming convention is applied to all Infrastructure Services defined in the Architecture

    Solutions Continuum (ASC) of the RAL. The same principle is used as DNS name, which is

    appended to the left to become more specific.

    Example:

    Level 1: IBS - Infrastructure Basic Service

    Level 2: Hosting-IBS Infrastructure Basic Service of type Hosting

    Level 3: Virtual-Hosting-IBS Infrastructure Basic Service of type Virtual Hosting

    Level 4: Windows-Virtual-Hosting-IBS Infrastructure Basic Service of type Windows Virtual

    Hosting

    Max 4 levels are defined and in total about 160 Infrastructure services are defined in the ASC.

    Each of these services should have corresponding Patterns which are in line with the defined

    standards in the company.

    When a new service is being created for one of Engie ITs clients, one or more services of the

    Architecture Solutions Continuum (ASC) are being instantiated and deployed in one of Engie ITs

    Data Centres.

    The infrastructure gets a name which will contain 2 parts:

    A Deployed Solutions Continuum (DSC) reference name something meaningful to the client

    and,

    The Architecture Solutions Continuum (ASC) reference, indicating the Type the Infrastructure

    Service according to the Architecture Solutions Continuum (ASC)

    3.2.2 RAL as a Modular Structure

    The RAL is a modular structure. The Services defined in the RAL represent a grouping of

    functionalities which are combined together to deliver a service which is useful and meaningful

    to Engie ITs clients.

    This modular structure exists in the Architecture Solutions Continuum as the different

    Infrastructure Basic Services (IBS), Infrastructure Platform Services (IPS) and Infrastructure

    Solution Services (ISS) have interdependencies.

    Example:

    Hosting-IBS always uses Housing-IBS because you need physical location to put a machine and

    then the Operating System Software can be installed to create a Hosting-IBS.

  • Applying the GNST to the Engie IT RAL

    14 | P a g e

    Figure 7: Dependencies in the ASC part of the RAL

    The modular structure exists as well in the Deployed Solutions Continuum (DSC) as it will

    combine different instantiations of Architecture Solutions Continuum (ASC) Services. The

    dependencies defined in the Architecture Solutions Continuum) (ASC) will reappear in the

    Deployed Solutions Continuum (DSC) they exist at design time and thus also during run time.

  • Applying the GNST to the Engie IT RAL

    15 | P a g e

    4 Artefact Requirements

    This chapter will elaborate on the Requirements of the Artefact

    The Artefact must be able to provide an answer to the question how an Infrastructure System

    will behave under change.

    Generalised Normalised Systems Theory (GNST) studies the impact of change on a modular

    structure. The input for the Artefact must thus be a modular structure representing an

    Infrastructure System.

    The Artefact must thus contain a mechanism to translate the Generalised Normalised Systems

    Theory (GNST) concepts toward an Infrastructure Context.

    Based on this translation, the 4 GNST can be tested against the modular Structure.

    The Artefact must contain a conclusion with regards the evolvability of the Infrastructure

    System and provide guidance to improve the evolvability of the Infrastructure System.

    Figure 8: Artefact Requirements

  • Applying the GNST to the Engie IT RAL

    16 | P a g e

    5 Artefact Design

    In the previous 2 chapter it has been defined what is being meant with Infrastructure and what

    the requirements are for an Artefact which will evaluate the behaviour of an Infrastructure

    System under change.

    This chapter will focus on the design of the Artefact. The Generalised Normalised Systems

    Theory (GNST) will be shortly introduced followed by the contextualisation of the GNST toward

    Infrastructure Systems. The chapter ends with the formal design of the Artefact.

    5.1 General info on the Generalised Normalised

    Systems Theory

    The Normalised Systems Theory (NST) is a theoretical framework which allows the investigation

    of Modular Structures under change and which is based on concepts of System Theory. The

    theory was first constructed and applied at the software level with focus on Modular Structures

    in Software Architecture. It provides an answer to the question on how to make software

    evolvable without degenerating the software and keeping it free of Combinatorial Effects.

    Combinatorial Effects (CE) are hidden coupling or dependencies in a system which increase with

    the size of the system.

    A system is considered unstable under change when the impact of a change is directly

    proportional to the change itself AND the size of the System.

    The NST has been applied at the Software Level10.

    The NST has been applied at the level of Industrial Automation Systems11.

    The NST has been applied at the level of Business Processes12.

    The NST has been applied at the level Enterprise Engineering13

    Based on the multiple applications of the NST outside its original Software scope, the theory got

    generalized in the General Normalized Systems Theory14.

    10

    Normalized Systems Herwing Mannaert, Jan Verelst - 2009 11 Towards adaptive flexibility in automation systems: industrial control software based on normalized systems theory - Dirk van der Linden 2014 12

    Towards designing Modular and Evolvable Business Processes. - Dieter VAN NUFFEL 2011 13

    On the Feasability of Normalized Enterprises: Applying Systems Theory to High-Level Design of Enterprises. - Philip HUYSMANS

  • Applying the GNST to the Engie IT RAL

    17 | P a g e

    Figure 9: Generalization of the Normalised Systems Theory - Peter De Bruyn

    A system is considered a Normalized System when the modular structure of the System is

    compliant with the following 4 principles:

    Separation of Concerns (SoC) A primitive should only contain one concern. 2 types of concerns are identified: change driver: each part of the modular structure which can independently change information unit: each part of the modular structure of which we are interested in contains independent information regarding its execution

    Separation of State (SoS) When a primitive uses another primitive as it is executed, both primitives should be separated by a state.

    Version Transparency (VT) A primitive which is used by other primitives should be version transparent

    Instance Traceability (IT) The input received to execute a particular primitive instance, as well as the output delivered by that primitive instance should be traceable to the particular version and instance of that primitive.

    Table 3: Normalization Principles

    When a System is compliant with these principles, it will have ZERO Combinatorial Effects (CE)

    under change which would make the system highly evolvable.

    14 Generalizing normalized systems theory: towards a foundational theory for enterprise engineering - De Bruyn Peter - 2014

    http://anet.uantwerpen.be/record/irua/opacirua/c:irua:120619

  • Applying the GNST to the Engie IT RAL

    18 | P a g e

    5.2 Contextualisation for Infrastructure Systems

    In order to investigate if a modular structure is compliant with the Generalised Normalised

    Systems Theory (GNST), the concepts of the GNST being Concern, State, Version and Instance

    must manifest themselves in the modular structure.

    For each module, a manifestation of Concern, State, Version and Instance must be found and

    then checked if those manifestations in the module respects the 4 Principles.

    For some Infrastructure Systems, State and Instance do not apply (see chapter 6).

    Example:

    A module representing Calculation

    Table 4: Manifestations of C, S, V and I

    A similar exercise must be made for each module which is part of the Infrastructure system

    under investigation:

    What are the manifestations of Concern, State, Version and Instance,

    Do those manifestations respect the 4 Principles

    If one or more of the principles are not respected, then CE will occur.

    5.3 Artefact Description

    5.3.1 Artefact input

    An ArchiMate model representing an Infrastructure System. The Infrastructure system is the

    aggregation of the different modules of which the Infrastructure System is made up.

    Dependencies between the modules are shown by means of the used by relation.

    5.3.2 Artefact Content

    1. GNST compliance check

  • Applying the GNST to the Engie IT RAL

    19 | P a g e

    For each module in the Infrastructure System, find the manifestations of Concern, State,

    Version and Instance. Then evaluate if the GNST principles are respected. If they are not

    respected, note the practical observed CE due to this non-compliance. Summarize the results in

    the table below:

    Table 5: GNST Compliancy check summary

    2. Evolvability of the Infrastructure System

    Based on the GNST compliance check, the Artefact will contain a general observation

    about the evolvability of the Infrastructure System.

    3. Guidance with regards to evolvability

    If the GNST principles are not respected, there will be issues related to future changes in

    the Infrastructure System. This section can contain guidelines on how to mitigate the

    impact or on how to restructure the Infrastructure System to eliminate or mitigate the

    impact.

  • Applying the GNST to the Engie IT RAL

    20 | P a g e

    6 Demonstrating the Artefact

    The Artefact will be demonstrated on 3 Infrastructure Systems defined in the Engie IT RAL:

    Housing-IBS

    Hosting-IBS

    Proxy-network-IBS

    The modular structures representing the Infrastructure Systems are not mirrors of the actual

    reality and contain simplifications. The objective is to demonstrate the Artefact can be applied,

    not to create imitation models.

    Before the Artefact is applied, the different modules of the Infrastructure System are

    introduced. The detailed study used to eventually fill out the GNST Compliancy check summary

    table, can be found in Appendix 1.

    6.1 Housing Infrastructure Basis Service

    Housing Infrastructure Basic Services (HousingIBS) offer basic Data Centre services to IT

    Equipment installed in a Data Centre, such as floor space, power, cooling, cabling.

    A Housing Infrastructure System is made up of the following modules:

    Location: Delivering a physical location, expressed in x, y coordinates in a physical room.

    Power: Delivering electrical power

    Cooling: Delivering cooled and conditioned air to IT equipment

    Rack: Delivering vertical stacking of equipment (z coordinate)

    Cabling: Delivering interconnectivity between IT equipment

    The 5 modules can be linked to each other in different topologies.

    Topology 1, referred to as Housing 1.0, represents housing of 20 years ago, but still applied in

    small data rooms.

    Topology 2, referred to as Housing 2.0, represents housing of 10 years ago, but still practiced in

    todays data centres

    Topology 3, referred to as Housing 3.0, represents the state of the art data centre setup.

    For each of the 3 topologies, the Artefact will be applied. The investigation of the

    manifestations of Concern, State, Version and Instance is identical in the 3 topologies and will

    be introduced before the Artefact is applied do the 3 topologies.

  • Applying the GNST to the Engie IT RAL

    21 | P a g e

    6.1.1 Manifestations of Concern, State, Version and Instance

    6.1.1.1 Location

    Concern: In this simplified model, location addresses 1 concern, delivery of x and y

    coordinates.

    State: Location has no manifestation of State. Location is a set of physical

    coordinates which do not require to persist their state externally nor

    exchange their state with other modules.

    Version: Location has no manifestation of Version. The model takes the assumption

    that the actual Location, does not change.

    Instance: Location has no manifestation of Instance. An equipment cannot be at 2

    different locations (2 different instances) at the same time.

    6.1.1.2 Power

    Concern: Power addresses multiple concerns. Delivery of electrical current to a specific

    location (physical cabling, copper, insulation, switch boards etc. making up

    the power grid) and connecting the IT equipment to the power grid.

    State: Power has a binary state on or off. By means of meters installed in the

    switchboards, colouring of the switches (on = red, off = green), led indicators,

    Power is able to persist its state outside of itself.

    Version: Power can have different versions. Electrical power can be delivered in

    different form (220V/50Hz, 3x380V/50Hz, 110V/60Hz) and connectors

    (outlets and connectors) can differ as well.

    Instance: Power has no manifestation of instance.

    6.1.1.3 Cooling

    Concern: Cooling addresses multiple concerns. Delivery of conditioned air, the

    distribution of the conditioned air to the right location and the extraction of

    hot air.

    State: Cooling has multiples states. The actual cooling installation will persist its

    state on a control board, indicating air temperature and humidity. Via sensors

    in the room, the distribution of the conditioned air and accumulation of hot

    air can be measured. The State of Cooling can be persisted outside the

    module, if the modules includes the necessary technologies (sensors and

    status indicators etc.) to do so.

  • Applying the GNST to the Engie IT RAL

    22 | P a g e

    Version: Cooling exists in different versions. Centralized, decentralized, air cooled,

    water cooled etc.

    Instance: Cooling has no manifestation of instance.

    6.1.1.4 Rack

    Concern: Rack can address multiple concerns, depending how it integrates with the

    other modules. Primarily Rack delivers vertical placement of IT Equipment a

    z coordinate. Rack can also serve as proxy for power, cooling and cabling. In

    the different Housing Topologies, the impact of using Rack for one or more

    concerns will be addressed.

    State: Rack has no manifestation of State.

    Version: Rack can have different versions. Racks come in different form factors.

    Instance: Rack has no manifestation of Instance.

    6.1.1.5 Cabling

    Concern: Cabling addresses multiple concerns. It acts a physical medium to transfer

    signals over a certain distance (via copper or fibre). It also connects IT

    equipment with this physical medium.

    State: The State of Cabling can be observed by means of control LEDs (if a signal

    detected yes/no, or the transfer speed of the data over the wire). These LEDs

    will be available on the IT equipment to which the cabling is connected.

    External components to Cabling make the State of Cabling visible.

    Version: Cabling can have different versions. The type of wire and insulation of the

    wire will determine maximum data transfer rates. The connectors of the wire

    come in different form factors.

    Instance: Cabling has no manifestation of instance.

  • Applying the GNST to the Engie IT RAL

    23 | P a g e

    6.1.2 Applying the Artefact to Housing 1.0

    6.1.2.1 Artefact Input Graphical Modular Structure

    Figure 10: Housing 1.0 Modular Structure

    The Housing 1.0 Pattern delivers the Housing-IBS service as described in the Engie IT

    Reference Architecture Library (RAL). This service is used by the IT Equipment.

    In Housing 1.0 Rack is only used to provide vertical space to the IT Equipment. The IT

    Equipment is directly linked to Rack, Power, Cabling and Cooling.

  • Applying the GNST to the Engie IT RAL

    24 | P a g e

    6.1.2.2 Applying the Artefact

    1. GNST Compliancy Check

    Figure 11: Effect of change to Housing 1.0

    Table 6: Housing 1.0 GNST Compliance summary

    (1): State concept reduced to persist state external to module

    4. Evolvability of the Infrastructure System

    The GNST principles are not respected and CE are expected under change. Observed CE confirm this hypothesis. Housing 1.0 is still used today in

    small data rooms. The effect of frequent change is best demonstrated in Figure 11. Housing 1.0 is not an evolvable system.

    5. Guidance with regards to evolvability

    Cleaner cabling, via cable trays may help reduce the entropy of the system. Use a Housing 2.0 topology.

  • Applying the GNST to the Engie IT RAL

    25 | P a g e

    6.1.3 Applying the Artefact to Housing 2.0

    6.1.3.1 Artefact Input Graphical Modular Structure

    Figure 12: Housing 2.0 Modular Structure

    In Housing 2.0, the IT Equipment is no longer directly linked with Cabling, Power and Cooling.

    Rack will contain proxies for:

    Cabling by means of integrated patch panels and cable trays.

    Power by means of Power Distribution Units.

    Cooling by means cooling fans at the top and bottom of the Rack.

    Rack becomes an Elements, as described in the Normalized System Theory (NST).

  • Applying the GNST to the Engie IT RAL

    26 | P a g e

    6.1.3.2 Applying the Artefact

    1. GNST Compliancy Check

    Figure 13: Power Distribution Units, Patch panels and cable trays

    Table 7: Housing 2.0 GNST Compliance summary

    (1): State concept reduced to persist state external to module

    2. Evolvability of the Infrastructure System

    Rack has the characteristics of an Element due to the proxies in Rack. Equipment can be added without introducing CE. When a Rack is full, a

    new Rack needs to be installed and at that point CE may be introduced again. CE are still present when changes occur in Cabling, Power and

    Cooling.

    3. Guidance with regards to evolvability

    Plan the layout of the computer room to anticipate the addition of new racks. Use a Housing 3.0 topology.

  • Applying the GNST to the Engie IT RAL

    27 | P a g e

    6.1.4 Applying the Artefact to Housing 3.0

    6.1.4.1 Artefact Input Graphical Modular Structure

    Figure 14: Housing 3.0 Modular Structure

    In Housing 3.0, a new module is introduced called Zone, which consists of 2 rows of Racks,

    installed back to back, with sufficient space between the 2 rows to allow human interventions.

    A Zone is a containerization concept, including all the Proxies for Power, Cooling and Cabling.

    More details on a Zone setup can be found in Appendix 1

  • Applying the GNST to the Engie IT RAL

    28 | P a g e

    6.1.4.2 Applying the Artefact

    1. GNST Compliancy Check

    Figure 15: Microsoft seaborne data centre container

    Table 8: Housing 3.0 GNST Compliance summary

    (1): State concept reduced to persist state external to module

    2. Evolvability of the Infrastructure System

    Zones can be stack next and onto each other, providing a mechanism to deliver housing facilities. The Version Transparency issues are solved by

    adding new containers containing version compliant components. Zones can be added theoretically indefinitely. Practically, there are always

    physical limits (for location, cabling, power and cooling) which will restrict this at some point.

    3. Guidance with regards to evolvability

    If zones can produce their cooling and power themselves, like the Microsoft seaborne data centre container (see Appendix 1 for more info).

  • Applying the GNST to the Engie IT RAL

    29 | P a g e

    6.2 Hosting Infrastructure Basis Service

    Hosting Infrastructure Basic Services deliver Hosting Services. Hosting is the combination of a

    hardware platform and an Operating System (OS), offering the possibility to exploit the

    hardware resources via an Operating System (OS). A Hosting Infrastructure System is made up

    of the following modules:

    Computer Hardware: Delivering process power, memory and Input/output control

    Operating System Kernel: Provides access to the Computer hardware

    Operating System Stacks: Different software stacks running on top of the OS kernel which

    allow applications to get access to the network and storage resources, access to the

    graphical power of the hardware, access to the OS process and resource

    management capabilities.

    Framework stacks: Different software stacks which allow the exploitation of the computer

    resources by combining and packaging the functionalities offered by the OS Kernel,

    Network, Storage, Graphical, Process and Resource Stacks in different software

    packages usable in different programming languages.

    The represented simplified Hosting Infrastructure System is OS agnostic. The Hosting

    Infrastructure System uses a Housing, Network and Storage Infrastructure System.

    The next sections will discuss 2 topologies:

    Hosting 1.0, a Physical hosting Topology

    Hosting 2.0, a Virtual hosting Topology

    For each of the topologies, the Artefact will be applied. The investigation of the manifestations

    of Concern, State, Version and Instance is identical in the 2 topologies and will be introduced

    before the Artefact is applied do the 2 topologies.

    6.2.1 Manifestations of Concern, State, Version and Instance

    6.2.1.1 Computer Hardware

    Concern: Computer Hardware addresses multiple concerns - CPU, GPU, MEM, IO. Computer

    Hardware is a fine grained modular structure and has been source of inspiration for

    the creation of the Normalised Systems Theory. The assumption is taken that,

    although multiple concerns are addressed, they can packaged in such a way that

    these multiple concerns are not source of CE.

    State: Computer Hardware has manifestation of State, the state is persisted outside the

    module and can be interrogated. Example is the ability to lookup CPU and memory

  • Applying the GNST to the Engie IT RAL

    30 | P a g e

    usage. Some Computer Hardware will also persist state to transmit information.

    Examples are the usage of buffers where one hardware module writes requests for

    instruction execution and another module will read those requests and execute

    them.

    Version: Computer Hardware has manifestation of Version - CPU versions, memory etc.

    Compatibility matrices will show which hardware versions are compatible and thus

    have Version Transparency.

    Instance: Computer Hardware has no manifestation of Instance.

    6.2.1.2 Operating System Kernel

    Concern: Operating System Kernel addresses multiple concerns. Pushing instructions to the

    CPU, transferring data from Memory to buffers, reacting on interrupts, performing

    Input/output The OS Kernel is an aggregation of functionalities.

    State: Operating System Kernels has a state and will persist this state in logs. Operating

    Systems Kernels work in a synchronous mode. State is not used to transfer

    information between sub modules.

    Version: Operating System Kernel come in different versions, linked to the Operating System

    release and Operating System patch level. Version compatibility issues can occur.

    Instance: Operating System Kernel has no manifestation of Instance.

    6.2.1.3 Operating System Stacks

    Concern: Operating System Stacks address multiple concerns. For instance in the Network

    stack, the 7 OSI layers are addressed to make sure data can be transferred from one

    computer to another.

    State: Operating System Stacks will persist their state in logs. Calling functions of an

    Operating System Stack happens synchronous, although the internal processing

    may be asynchronous.

    Version: Operating System Stacks come in different versions, linked to the Operating System

    release and Operating System patch level. Version compatibility issues can occur.

    Instance: Operating System Stacks have no manifestation of instance

  • Applying the GNST to the Engie IT RAL

    31 | P a g e

    6.2.1.4 Framework stacks

    Concern: Framework stacks address multiple concerns. They can contain all kinds of

    combinations of calls toward the Operation System Stacks and the Operating

    System Kernel.

    State: Frameworks may or may not persist their state. This will fully depend on the

    Framework.

    Version: Frameworks come in different versions, linked to the Operating System release and

    patch level. Version Compatibility issues can occur.

    Instance: Multiple instances of Frameworks may run on the OS.

    6.2.2 Applying the Artefact to Hosting 1.0

    6.2.2.1 Artefact Input Graphical Modular Structure

    Figure 16: Hosting 1.0 modular Structure

  • Applying the GNST to the Engie IT RAL

    32 | P a g e

    6.2.2.2 Applying the Artefact

    1. GNST Compliancy Check

    Table 9: Hosting 1.0 GNST Compliance summary

    2. Evolvability of the Infrastructure System

    Hosting 1.0 has serious evolvability issues mainly due to non compliance with Version Transparency. The CE are unavoidable unless OS versions

    and Commercial Of the Shelf (COTS) Applications would respect the 4 principles.

    3. Guidance with regards to evolvability

    The rule One Operation System, One Application applies.

  • Applying the GNST to the Engie IT RAL

    33 | P a g e

    6.2.3 Applying the Artefact to Hosting 2.0

    6.2.3.1 Artefact Input Graphical Modular Structure

    Figure 17: Hosting 2.0 modular Structure

    Hosting 2.0 contains a new module the Hypervisor. A Hypervisor is a special Operating

    System. It allows the installation of multiple Operating Systems on top of the

    Hypervisor. The Hypervisor virtualizes the Hardware, transforming actual CPUs into

    virtual CPUs. The main driver for virtualisation is better use of the hardware resources.

  • Applying the GNST to the Engie IT RAL

    34 | P a g e

    6.2.3.2 Artefact

    1. GNST Compliancy Check

    Table 10: Hosting 2.0 GNST Compliance summary

    2. Evolvability of the Infrastructure System

    Operating System Virtualisation does not solve the evolvability issues found in Hosting 1.0. Hosting 2.0 has the same limitations.

    3. Guidance with regards to evolvability

    The rule One Application, one Operating System stays applicable but with the advantage that multiple Operating Systems can run on one

    machine.

    The usage of Application Virtualisation techniques can solve Version Transparency issues (see Appendix 2 for more information).

  • Applying the GNST to the Engie IT RAL

    35 | P a g e

    6.3 Proxy Infrastructure Basis Service

    Proxy Network Infrastructure Basic Services offer the functionality to convert N to M outbound

    traffic into 1 to M outbound traffic as all outbound N are hidden behind the 1 proxy. It is also

    caches information, applies network translation and checks if traffic is authorised based on

    allowed targets and Identity information provided by the proxy user. A Proxy Pattern is made

    up of the following modules:

    Proxy Engine: Performs the Proxy activity based on input from the other modules.

    NAT Engine: Performs the Network Address Translation to hide the actual IP of Source and

    Target.

    Rule Engine: Checks if based on Authorised Targets and Identity information provided by the

    Source, traffic is allowed of not.

    Caching: Caches information of the Target to optimize traffic.

    Authorized Targets: Manages the list of Authorised Targets.

    Logging: Logs information on the traffic exchanged between Source and Target.

    Proxy controls traffic from the company network towards the Internet. For security reasons, the

    usage of a proxy for all Sources connecting to Targets on the Internet, is considered mandatory.

    But not all Sources are able to connect to a Target via a Proxy! To study the impact of this, the

    Extended Proxy Pattern is introduced, which includes the Source as well.

    6.3.1 Manifestations of Concern, State, Version and Instance

    6.3.1.1 Proxy Engine

    Concern: Connectivity between Source and Target.

    State: Indirectly persists its state in the Proxy log (who accessed what when). Traffic

    between Source and Target is synchronous, so no persistence of exchanged

    information.

    Version: Proxy Engine is part of the Proxy Pattern. The Version of the Engine will be the

    version of the Pattern.

    Instance: There is no manifestation of Instance in the Proxy Pattern as a Source will

    communicate via one Proxy to its Targets.

  • Applying the GNST to the Engie IT RAL

    36 | P a g e

    6.3.1.2 NAT Engine

    Concern: Network Address Translation of Source and/or Target.

    State: No manifestation of State.

    Version: NAT Engine is part of the Proxy Pattern. The Version of the Engine will be the

    version of the Pattern.

    Instance: There is no manifestation of Instance in the Proxy Pattern as a Source will

    communicate via one Proxy to its Targets.

    6.3.1.3 Rule Engine

    Concern: Checks compliance with identity and authorised target

    State: No manifestation of State

    Version: Rule Engine is part of the Proxy Pattern. The Version of the Engine will be the

    version of the Pattern.

    Instance: There is no manifestation of Instance in the Proxy Pattern as a Source will

    communicate via one Proxy to its Targets.

    6.3.1.4 Caching

    Concern: Cache information of the Target to avoid unnecessary traffic.

    State: No manifestation of State

    Version: Caching is part of the Proxy Pattern. The Version of the Engine will be the

    version of the Pattern.

    Instance: There is no manifestation of Instance in the Proxy Pattern as a Source will

    communicate via one Proxy to its Targets.

    6.3.1.5 Authorized Targets

    Concern: Manages the list of allowed Targets

    State: No manifestation of State

    Version: Authorized Targets is part of the Proxy Pattern. The Version of the Engine will

    be the version of the Pattern.

    Instance: There is no manifestation of Instance in the Proxy Pattern as a Source will

    communicate via one Proxy to its Targets.

    6.3.1.6 Logging

    Concern: Logs which source has connected to which target, when and how.

    State: No manifestation of State

  • Applying the GNST to the Engie IT RAL

    37 | P a g e

    Version: Logging is part of the Proxy Pattern. The Version of the Engine will be the

    version of the Pattern.

    Instance: There is no manifestation of Instance in the Proxy Pattern as a Source will

    communicate via one Proxy to its Targets.

    6.3.1.7 Source

    Concern: Exchange information with Target.

    State: No manifestation of State.

    Version: Source can have different versions. Different version of browsers and

    different versions of applications.

    Instance: Of a specific browser and of a specific applications, multiple instances can be

    active at the same time. Each needs to be correctly configured to work with

    the Proxy.

    6.3.2 Applying the Artefact to Proxy

    6.3.2.1 Artefact Input Graphical Modular Structure

    Figure 18: Proxy modular Structure

  • Applying the GNST to the Engie IT RAL

    38 | P a g e

    6.3.2.2 Artefact

    1. GNST Compliancy Check

    Table 11: Proxy GNST Compliance summary

    2. Evolvability of the Infrastructure System

    The Proxy Pattern respects the GNST principles and is highly evolvable. However, the Extended Proxy Pattern, being the default usage of any

    source to use a Proxy to connect to a Target on the Internet, does not respect the GNST. CE can be expected and CE are observed.

    3. Guidance with regards to evolvability

    Make sure all sources are Proxy compatible. Use tooling to distribute proxy configuration items to all sources (see Appendix 3 for more

    information)

  • Applying the GNST to the Engie IT RAL

    39 | P a g e

    7 Evaluation of the Artefact qualitative analyses

    The analysis of the Infrastructure Modular Structures was presented to the Expert Team in the form of Appendix 1. The summary of

    the analyses has been presented in Chapter 6. The Expert Team was requested to score 3 aspects:

    The correctness of the Infrastructure Modular Structure the Artefact input

    The correctness of the Artefact Analyses Manifestations of Concern, State, Version, Instance and observed CE.

    Did the Artefact analyses provide extra insights?

    Correctness of the Infrastructure Modular

    Structure

    Correctness of the Artefact Analyses Did the Artefact analyses provide extra

    insights?

    Scoring and scale:

    1 - The model has no link with reality

    2 - The model lacks key elements of reality

    3 - The model contains some key elements of reality

    4 - The model contains most key elements of reality

    5 - The model is an imitation of reality

    Scoring and scale:

    1 Analyses in contradiction with reality

    2 - Analyses lacks key elements of reality

    3 - Analyses contains some key elements of reality

    4 Analyses contains most key elements of reality

    5 - Analyses is fully aligned with reality

    Scoring and scale:

    1 Not at all

    2 - Minor confirmation current insights

    3 - Partially confirms current insights

    4 - Confirms current insights

    5 - New insight

    Average: 4.2/5 Average: 4.3/5 Average: 4.4/5

    Table 12: Expert Team Evaluation 11 of the 13 Expert Team members

  • Applying the GNST to the Engie IT RAL

    40 | P a g e

    8 Evaluation of the Artefact feedback

    The Expert Team did not limit itself to scoring the Artefact but provided additional information

    with regards to the value of the Artefact and observed CE which they could link to the a

    violation of the Generalized Normalised Systems Theory (GNST) Principles.

    8.1 Value of the Artefact

    A member of the Expert team considered the GSNT Principles as Input for the creation of

    Architecture Principles

    Two other members find it to be a Meta model which is in line with practical decision making

    and Theoretical model explaining why Infrastructure is so tricky.

    8.2 Observed CE

    The Infrastructure Modular Structures analysed in Chapter 6 are all part of the Infrastructure

    Basis Services (IBS) layer of the Reference Architecture Library (RAL). Expert Team members

    provided input to practically observed CE which can be linked to a violation of one of the 4

    GNST principles, for the 2 other layers Infrastructure Platform Services and Infrastructure

    Solution Services. Table 13 contains some CE, observed at the other layers, and classified

    according to the Principle they violate. The full details of the analyses can be found in Appendix

    2.

    The table demonstrates that CE are to be found in all aspects of Infrastructure both physical

    and software and that a detailed analyses could be made of those CE by means of the Artefact.

  • Applying the GNST to the Engie IT RAL

    41 | P a g e

    15

    Table 13: CE at different layers of the RAL

    15

    Data Guard Support for Heterogeneous Primary and Physical Standbys in Same Data Guard Configuration (Doc ID 413484.1)

  • Applying the GNST to the Engie IT RAL

    42 | P a g e

    9 Artefact Value

    Chapter 6 to 8 focused on demonstrating and scoring the Artefact. The Expert Team already

    provided some feedback with regards to the value of applying the Artefact. In this chapter, the

    value of the Artefact will be studied from the point of view of the existing Engie IT Reference

    Architecture Library (RAL), the value the Artefact has for an Architect and the value for the

    Generalized Normalized Systems Theory (GNST).

    9.1 Value for RAL Engie IT

    The Architecture Solutions Continuum part of the RAL (ASC) contains the Infrastructure Building

    Blocks Engie IT uses to construct Infrastructure Services for its customer.

    Each of the building blocks is a grouping of functionality and these groupings correspond to real

    life building blocks which are available on the market. Infrastructure Building Blocks are driven

    by what is available at the market. They deliver a certain grouping of functionality and there is

    no control over how they are delivered and realized. Essentially there are black boxes.

    The models studied in Chapter 6 show generic models of such black boxes. Applying the

    Artefact to these models creates insight with regards to the evolvability. Sometimes evolvability

    of the infrastructure is not required. In such case applying the Artefact has no value. But as put

    forward in the Problem Statement, an Agile Business required an Agile and thus evolvable IT

    Infrastructure. Analysing the IT Infrastructure with the Artefact will provide clarity with regards

    to the limits of the evolvability of the IT Infrastructure Systems put in place or IT Infrastructure

    Systems which the company plans to put in place.

    Creating models which focus on the modular structure of Infrastructure Systems is a necessity

    to apply the Artefact. Making such models for Infrastructure Systems which require a high

    degree of evolvability, is valuable. It allows to apply the Artefact and make theoretically and

    practically sound predictions about the evolvability of an Infrastructure System. Infrastructure

    Systems can thus be build or purchased with evolvability in mind and the limitations are known

    up front.

    9.2 Value for the Architect

    Architects are often requested to create roadmaps. A roadmap is all about the evolution of a

    system. The usage of the Artefact will help the Architect making scientifically correct

    statements about the evolvability of an IT System. The Artefact should be the mandatory tool

  • Applying the GNST to the Engie IT RAL

    43 | P a g e

    for such types of analyses. The tool becomes for the Architect what a Risk Taxonomy is for a

    Risk Manager, or Statistics and Data Mining Algorithms are for a Data Scientist.

    The ex-ante proven evolvability issues can be used as input for:

    Architectural Risk Analysis

    Triggering mitigation actions

    o Close follow up of system usage and evolution

    o Provide tooling to temper the CE (see Appendix 3)

    IT budget creation or IT investment planning

    9.3 Value for the Generalized Normalised Systems

    Theory GNST

    The demonstration of the Artefact in Chapter 6 shows that the GNST can be applied on modular

    structures representing IT Infrastructure. The theory has been made applicable and practical in

    a new field. It shows the General nature of the theory and contributes to the further

    proliferation of the theory in other parts of IT and science.

  • Applying the GNST to the Engie IT RAL

    44 | P a g e

    10 Conclusions and Reflections

    Recalling the Research Question:

    Can the Generalised Normalised Systems Theory (GNST) be used to study the behaviour of IT

    Infrastructures under change?

    This leads to the following research questions:

    Part 1: Can the Generalised Normalised Systems Theory (GNST) principles be applied to

    modular structures representing IT Infrastructure?

    Part 2: Can Combinatorial Effects (CE) - hidden coupling or dependencies in a system which

    increase with the size of the system - be predicted and recognized at the IT Infrastructure level

    by using the Generalised Normalised Systems Theory (GNST)?

    Part 3: What is the added value of using the Generalised Normalised Systems Theory (GNST)

    at the IT Infrastructure level?

    10.1 Conclusions

    Using Design Science Research Methodology, an Artefact has been designed (Chapter 5),

    demonstrated (Chapter 0) and evaluated (Chapter 7) which is used to test the Generalized

    Normalised Systems Theory on a modular structure representing an Infrastructure System. The

    Engie IT Reference Architecture Library has been used as source for Infrastructure Systems. The

    construction of the Artefact has been validated and evaluated (Chapters 7 and 8) by an Expert

    Team and the value of the Artefact has been discussed (Chapter 9)

    The successful usage of the Artefact leads to the conclusion that:

    1. The Generalised Normalised Systems Theory (GNST) principles can be applied to modular

    structures representing IT Infrastructure.

    2. CE can be predicted in IT Infrastructures using the GNST principles. Predicted CE can be recognized

    in IT Infrastructure Systems.

    3. Using the GNST at IT Infrastructure level provides valuable insight into the evolvability of IT

    Infrastructure Systems. It can be used to analyse:

    i. Existing IT Infrastructure Systems from which evolvability is expected.

    ii. The design of IT Infrastructure Systems (to be build or to be purchased) and predict the

    evolvability of the System.

    iii. Create Roadmaps of IT Infrastructure System

  • Applying the GNST to the Engie IT RAL

    45 | P a g e

    10.2 Reflections

    The subject of this research has led to a number of reflections which do not answer the

    Research Question but who do merit to be mentioned or should be subject of further

    investigated.

    10.2.1 GNST knowledge as essential Architecture Skill

    The GNST can be applied on any modular structure and most system can be represented by a

    modular structure. The trick is to find the correct manifestations of Concern, State, Version and

    Instance. Once these manifestations are known, the presented Artefact can be used to analyse

    the behaviour of the modular structure under change.

    Knowing and applying this is an essential skill for an Architect.

    Within Engie IT, for some years now, Infrastructure Architects are being trained to model in

    ArchiMate and to understand the usage of views and viewpoint. Knowing and applying the

    GNST concept is the next skill which needs to be taught to all Architects and applying the

    Artefact should become mandatory for every Architecture of an IT Systems which is bound to

    evolve.

    10.2.2 Artefact evolution

    Infrastructure Systems are packed with CE they are all over the place. But are all CE equally

    harmful? Appendix 3 talks about tooling which help to reduce the effect of CE or who will

    automate the resolution of CE.

    The Artefact should be further refined to make a distinction between benign CE and harmful

    CE. A link with Architecture Risk and risk mitigation should be made as well.

    The refinement the Artefact could be a subject for a new Master Thesis.

    10.2.3 Roadmaps and Lehmans Law

    An important trigger for the development of the Normalized Systems Theory is Lehmans Law

    of increasing complexity and degeneration of software due to change.

    Verelst and Mannaert16 argue based on this law that:

    If the cost of a software change at T0, is X0

    16

    Normalized Systems Herwig Mannaert, Jan Verelst - 2009

  • Applying the GNST to the Engie IT RAL

    46 | P a g e

    Then the cost of the same change at T1 will be X1

    Where T1 > T0 and X1>X0

    Cost of change increases over time.

    What if the same reasoning is applied at IT Infrastructure level with regards to changes on IT

    Infrastructure components?

    Would it be better, from a financial point of view, to upgrade as soon as possible when new

    version of a Software Infrastructure Component is released, or to wait till the existing

    components are out of support?

    Would it be better to invest if a Zone concept in a Data Centre (according to les rgles de lart

    as discussed in 6.1.4), or will the sum of all incremental costs be lower (or higher) than the

    initial up-front investment?

    Which elements need to be part of such a business cases and how could the degeneration

    effect described by Lehman, be factored in.

    Trying to answer those questions could be a subject for a new Master Thesis.

    10.2.4 Configuration is the new Code

    Infrastructure Systems are customized by means of setting parameters and configuration

    settings. The total sum of all these configuration is as valuable as the source code of in house

    developed applications. The way Infrastructure Systems are configured has impact on the

    evolvability of the Infrastructure Systems, similar to the way code is written has impact on the

    evolvability of an application.

    The next major technological shift at the Infrastructure Layer will be the Software Defined

    Network, Software Defined Data Centre and Software Defined Infrastructure.

    Structuring policies and configuration will be essential for proper functioning and the

    evolvability of the IT Systems.

    Can the N