Ram

Document Sample
Ram Powered By Docstoc
					       Service Engineering:
Different Perspectives on Services
     and their Relationships to
          Infrastructures
       C. V. Ramamoorthy, PhD
       UC Berkeley, CS Division
       ram@eecs.berkeley.edu
                       Overview

   Our approach and emphasis
       Ancient Chinese Saying (Wen Jinhno, Chinese
        Prime Minister)
          “For a Grander Sight, Climb to a Greater Height”
          It’s not about tourism. Higher the level of
           abstraction, farther will be reach of a concept –
           other disciplines and applications




                         C.V. Ramamoorthy, UC Berkeley
                         Goals - Objectives
   Survey of Service Technology
    - Primarily Information-Communication Technology (ICT) based
   Its Evolution
   Definitions – General, Current, as Practiced
    •   SOA, SAAS, etc.
    •   Carly Fiorina
   Research Topics
    •   Application Focus
         •   Impact of Moore & Metcalfe’s Law
         •   Application Categorization based on
              •   Infrastructural aspects
              •   Internet – Web aspects
              •   Core – Competency aspect


                                  C.V. Ramamoorthy, UC Berkeley
Automatability Discovering Service
Patterns suitable for Customization
   Service Composition & Decomposition
   Service Components & Standardization
   Service Function Specification
   Schematic Capture & Analysis
   Evaluation
   Parallelizability



                  C.V. Ramamoorthy, UC Berkeley
                    Continued…..
   Clustering
   Critical Service Activity Patterns
   Ideas & Concepts
    •   Mutual Attraction of Service Components
    •   Separation of Concerns
    •   Constraint based Separation




                         C.V. Ramamoorthy, UC Berkeley
      Execution of Service Activities
   Important but often ignored
   Ideas from Gen. Patton, W.Buffett, S.Covey,
    & M.Gladwell
   Phases in Execution
   Models
   Ideas




                     C.V. Ramamoorthy, UC Berkeley
                    Visions
   Internet of the Mind (KOZMETSKY)
   Pervasive Technology (Y. Mizuno)
   Service Ecology
   Information-Communication Technology




                    C.V. Ramamoorthy, UC Berkeley
               Service Definition
   Dictionary Definition
       An organized System of appliances,
    products, personnel, and other resources to
    supply activities needed to satisfy a public or
    private need.




                       C.V. Ramamoorthy, UC Berkeley
                    In Summary
   System design in T.M. is properly capturing
    customer requirements and developing concepts
    and design architecture based on customer
    needs
   Parameter design is analyzing and identifying
    essential functional and nonfunctional
    characteristics (parameters) and rearranging
    them to reduce application complexity.
       It emphasizes dependable performance from
        customer viewpoint despite expected hazards

                         C.V. Ramamoorthy, UC Berkeley
   Knowledge Utilization and Technology Transfer

Impact




                       Elec                           Digital
   Water    Steam      Chem                   Petro   IT        Bio-
   Power               Civil                  Chem    Comm      Genetics
   60 yrs   55 yrs    50 yrs               40 yrs     30 yrs    Time


                     C.V. Ramamoorthy, UC Berkeley
Growth                                                         W.Corrigan
               Knowledge                                       Extended
                                    Technology
                                                     Tools
                                                             Implementation




   Latencies
   X1<X2<X3




                     X1            X2               X3
                                                                      Time
                    C.V. Ramamoorthy, UC Berkeley
Growth
                 Knowledge
                          Technology
                             Tools
                                Implementation
                                                     Compression and
                                                     Convergence



   Latencies
   X1<X2<X3




                       X1X2 X3
                                                                       Time

         Phases in Technology Transfer
                     C.V. Ramamoorthy, UC Berkeley
Growth                Growth or Decline

             Manual effort




             Service industry

                                               Automation


                                                                  Time
    60 yrs   55 yrs      50 yrs              40 yrs      30 yrs
                         C.V. Ramamoorthy, UC Berkeley
Service Industry Functions


          Manual                           Intellectual
 Manual     +                                   &
          Mental                          Book-keeping


          Evolution




          C.V. Ramamoorthy, UC Berkeley
         Software as a Product
   E.g. MS-Windows
    Buy the CD-ROM.
    MS-Windows controls many application
    systems. When MS upgrades one thing
    you have to change everything else and
    you pay for the up-date fee.
    (Not a multi-tenant shared system – Marc
    Benioff, Salesforce.com)

                   C.V. Ramamoorthy, UC Berkeley
      Software as a Service (SAAS)
   E.g. Google, eBay, Yahoo, Aamazon.com,
    salesforce.com
   Online updating and maintenance of the Web-
    based System, transparent to the user
   The application service is maintained by the
    Service Provider
   Analogy to using Taxi Cabs



                     C.V. Ramamoorthy, UC Berkeley
     Service Oriented Architecture
Functional procedures (software programs) are broken
  down into standard components that can be
  reassembled to meet specific needs

               Carly Fiorina - Formerly of H.P.
               (Duality)
                 Need
    Systems                                              To
                         Services                        Satisfy
    Products                                                       Public/Private
                                                                   needs
                 Need



                         C.V. Ramamoorthy, UC Berkeley
      Multi-Tenant Shared Architecture
             E.g. Salesforce.com
   Uses the Concepts of Google, Yahoo, etc.
    (Consumer Internet Companies) & moves them
    to the business environment (Marc Benioff,
    Salesforce.com)
   Idea: The portal-website on which the service
    components reside are like town houses &
    apartments, occupied by multiple tenants – of
    shops, offices and residences. It functions like a
    shopping mall or market.
                       C.V. Ramamoorthy, UC Berkeley
                  Continued…
   Many mini-Googles set up their standardized
    service component stores and servicing, leasing
    of components to service shoppers &
    developers.
   They also provide guaranteed online updating
    & maintenance for their software products. The
    portal-website is the shopping mall.



                      C.V. Ramamoorthy, UC Berkeley
Microsoft and Google Strategies
   Microsoft (MS) (Product Company)
       Provides service by selling products “Windows Vista”
        has experienced great delays. It will be introduced
        next year (2007)
       From now on MS will follow new strategy (Google
        strategy)
       Traditionally, MS bunched up all changes (big and
        small) into one big patch and made a release
       Big changes by a single release confuse costumers


                          C.V. Ramamoorthy, UC Berkeley
                     Continued

   Google (Service Company)
       Small changes, updates frequent
       Introduce big changes incrementally
       Sustain customer habits for a long time
       All maintenance changes invisible to users
       Provides service directly through the Internet



                       C.V. Ramamoorthy, UC Berkeley
Microsoft:
MS Products are sold on CD-ROMs.
Service is requested by the user
& MS may provide the service


Google:
Service is provided by Google.
User never sees the products.


                C.V. Ramamoorthy, UC Berkeley
       MS-Practices – Currently
   Trend towards LARGER component sizes
   Fewer subsystems per product
   Too many components could create
    assumption conflicts and interface
    discrepancies
   Also smaller components and too many of
    them in system makes the system difficult
    to manage

                   C.V. Ramamoorthy, UC Berkeley
         MS Practices – Continued
   Conclusion: Advantage of using service strategy
   Remember
       ICT management & maintenance are services
       IBM has become a service-provider company
   Trend: The more complex the function, they ask
    some service provider to manage it. Don’t like to
    worry about maintaining complex ICT systems.
    Outsource management and maintenance.


                        C.V. Ramamoorthy, UC Berkeley
  High complexity due                                   High complexity due
    to multi-function                                      to many single-
      components                                        function components




                  Low complexity



(From Design of Trustworthy Software – Jaiswal & Patton)




                        C.V. Ramamoorthy, UC Berkeley
   Highly developed countries such as USA
    are becoming service-oriented societies
   By the yeah 2010
       80% of all workers in the US will be in the
        service industry –Newsweek
   Dictionary definition of service
       Organized system of products, appliances,
        and employees supplying activities and
        accommodations needed by public, e.g. TV
        repair service, message service, etc.

                       C.V. Ramamoorthy, UC Berkeley
   Software drives information technology (IT)
   Needs Driven
       Personal --Humanistic, and Individual
       Professional –Customized functions
       Societal (Community) –Altruistic
   Knowledge Intensive
       Clients, Providers, Developers
   Human Interaction Intensive
       Collaboration, Co-Operation, Learning, Teaching,
        Innovating
   Service Industry Functions
       Advance with Knowledge Growth, Technology
        Advances and Societal trends

                          C.V. Ramamoorthy, UC Berkeley
          Manufacturing      Services             Agriculture and mining


     90
     80
     70
     60
     50
     40
     30
     20
     10
      0
          1810      50              1900                 50         99
          year
                          Source : The Economist, December 31st 1999


The Growth of Service Industry Functions in US
                             C.V. Ramamoorthy, UC Berkeley
    Distinguishing Characteristics of
            Service Industry
   Information Technology Driven: software
    is the primary ingredient
   Knowledge Intensive
   Human Interaction Intensive
   Team-based Functions
   Service Industry Functions


                   C.V. Ramamoorthy, UC Berkeley
   The Highly Developed Countries are Moving
    Away from
       Industrial and Manufacturing Economy towards a
        Service Oriented Economy
   Examples of Service Industries
       Finance
       Travel
       Telephone and Communication
       Utilities
       Health Care
       Entertainment
       Etc.

                         C.V. Ramamoorthy, UC Berkeley
   Service Industry
       Almost Total Dependence on I.T.
       Mostly Software Engineering Discipline
   Similarities between Software Industry and
    Service Industry
       Knowledge Dependency
       Collaborative Aspects
            Complexity –One person cannot do it all
            Intellectual Intensity
            Book Keeping (Details) –Data collection
       Flexibility in Dealing with
            Variability in Interactions
                  Man-man, man-machine, team-team, team-machine
            Decisions and Choices
            Facilitating the Goal the Client Wants to Achieve

                                C.V. Ramamoorthy, UC Berkeley
            Application Classification
   These are overlapping services Applications
    •   General Service Applications
           Postal Service, FedEx, Maintenance,
           Utilities
    •   IT-Web Based
           Information-Communication
           Technology (ICT) based
           E.g. ICT as a Cottage Industry
           for Developing Countries
                   SOA -Based
                         C.V. Ramamoorthy, UC Berkeley
                          Continued…
   Core-Competency Based
    •   ICT-Core
            MS, IBM, Oracle, SAS,
           Google
    •   Non ICT-Core
         •   Supply Chain Support Services
         •   Transportation System
         •   Air Traffic Control System
    •   Infrastructure Based Services
         •   Power Supply & Distribution
         •   Water Management
                               C.V. Ramamoorthy, UC Berkeley
           Internet-Web based Service
             Application Classification
                   (overlapping)
   Application Service Provides
    •   Social Networks
    •   Personal Content based
    •   Search Engines
    •   Context based
    •   Network based Audio-Video Service
               Youtube (Google)
               Grouper (Sony)
               Yahoo-Jumpcut (These help in editing Video
               clips)    C.V. Ramamoorthy, UC Berkeley
                  Continued…
•    Network traffic analyzers (frequency, QOS, etc.)
•    Sharing Sites
•    Editing Sites-Supply tools
•    Blogs and personal content info
    are carefully analyzed to discover habits &
     preferences of users & later used by enterprises
    to target personalized ads (Amazon.com)




                       C.V. Ramamoorthy, UC Berkeley
        Needs-Knowledge – i/s Chain
Human                         Product                                    Tools &     Knowledge
                Services                              Technology
Needs                         Systems                                    Education   R&D
                              Processes




 i/s               i/s             i/s                        i/s           i/s         i/s

        •   Adaptive, Dynamic, & Changing
        •   We know “what” drives “what”
        •   We know what’s ahead of us (future?)
            and what’s with us & behind us (Present & Past)



                                         C.V. Ramamoorthy, UC Berkeley
Physical Infrastructure –
  Personalization Tree
                                                Personalization by user
                                                Customization

       Humanization


         Front End


                                                    Back
        Middleware
                                                    End

       Physical Layer
    Internet Connections
   Infrastructural Support

(A VARIATION OF ISO MODEL)
                C.V. Ramamoorthy, UC Berkeley
    Network Infrastructural Growth & Its
    Impact on Service Application Growth
   Models of Growth based on Infrastructures
    A. Brasilica Model – (Kubitschek-LeCorbusier)
    1956
    •   Creation of the capital of Brazil. City was in the
        middle of Amazon jungles & also at the middle of
        nation. Because of its importance, a strong
        transportation system was needed. The
        transportation system funneled industrial &
        agricultural growth
    •   Message: Infrastructures & the dependent
        application grow to satisfy needs.
                          C.V. Ramamoorthy, UC Berkeley
                     Continued…
   B. Roosevelt-New Deal Model (1930’s
    Depression Era)
    •   Roads & transportations (Infrastructures) were
        created across U.S. to support employment &
        funnel economic growth.
    •   New transportation Systems helped auto industry
        (Henry Ford) & the rapid growth of cities &
        suburbs. Nation’s economy grew.
    •   Message: Needs create the growth of
        Infrastructures.

                         C.V. Ramamoorthy, UC Berkeley
     C. The “Net” – Effect due to Moore
              & Metcalf’s Law
   Moore & Metcalf Laws synergize each other –
    (Push-Pull effect)
   Create an exponential growth subject to
    physical limits & natural saturation
      Herbert Simon’s Emergent Process
       (e.g. 1+1 >> 2)
   Also called the Rising Tide effect “Rising Tide
    lifts all the boats” – A variation & Result
    Net-Effect lifted Google, Yahoo, MS.,…
                       C.V. Ramamoorthy, UC Berkeley
Moore’s Law
     Component density & performance of semi-
      conductor chips double every 18-24 months with
      the cost remaining the same
     Cheap components created small inexpensive but
      powerful Commodities called PCs & mobile phones.
      They have become part of our day-to-day living
      infrastructure.
Metcalf’s Law
     Value (usage & growth) of a network is proportional
      to the square of the number of its users (n). Value
      (usefulness & growth) of a network increases
      “exponentially” with the extent of its use
      (applications) & the number of its users
                        C.V. Ramamoorthy, UC Berkeley
                  Continued…
•   Disagreements – (Odlyzko & Tilly at Univ. of
    Minnesota) ~ Growth is proportional to n.log n
•   “Reed’s Law” says Metcalf’s Law
    underestimated the network value
•   Metcalf’s law explains the growth of “Social
    Networks”.
    E.g. MySpace (FOX), Yahoo360, Facebook,
    Linkedin, etc.


                      C.V. Ramamoorthy, UC Berkeley
      Infrastructures & Applications
•   Many Web-Services are based on
    Infrastructural Business (Value)
    Growth
    E.g. client Growth due to Personal Content Applications


                        Google




         Myspace                      Youtube
                (Fox)

       Warren Buffet: “I am the flower.
       You are the watering cans.”

       Myspace = a place for friends
                                 C.V. Ramamoorthy, UC Berkeley
     Exploring the opportunities for
    automated Executions in Service
              Applications
•   A Summary
•   Meaning of Automated Executions.
•   Service Applications – Requirements & Specification.
•   Service Element Classification (Writer’s Paper of
    2000. IEJCE Trans.Comm. May 2000 on Service
    Industry)
•   Lack of Representation methods for Service
    Sequences (Standard Forms)


                      C.V. Ramamoorthy, UC Berkeley
                  Continued…
•   Given the Service Activity Sequence, its schematic
    capture in analyzable representation. Activity
    Decomposition.
•   Discovering of un-executable, Redundancies,
    dangerous & Critical Sequences.
•   Discovering Parallel executable sequences under
    time, Space, & resource constraints
•   Strategy & Planning (Mintzberg)
•   Exceptional Conditions & Options


                      C.V. Ramamoorthy, UC Berkeley
                      Continued…
•   Some preliminary observations & Results
     •   Parallel Executions       Difficult to find necessary &
         Sufficient Conditions like Bernstein’s conditions in
         Program executions
     •   Parallel Executions provide more options & opportunities
         but synchronizations pose high overheads in time &
         resources.
     •   Parallel Executions are beneficial if the threads
         (sequences) are long.




                           C.V. Ramamoorthy, UC Berkeley
                Some results so far
•   Developed methods for identification
•   Good candidate activity sequences for automated
    executions
     •   This is important in those service applications which
         involve physical (manual) activities.
     •   We can show that the rules for parallel execution & the
         rules for Component Clustering are “Complementary”.
         E.g. If you cannot break them into parallel execution
         sequences, you can cluster them & consider them for
         automated execution
     •   Need to consider Gomaa’s type of pattern analysis &
         ideas of data-mining for this purpose

                           C.V. Ramamoorthy, UC Berkeley
               Service Execution
   Most Important & generally neglected
   The essence & the meat of Service
   Gen. Patton’s Quote
   Warren Buffett - Focus & Concentrate
   Stephen Covey – Always keep the end in mind
   Malcolm Gladwell – “Blink”
   Gorbachev : “Execution is the most difficult
    part. It’s like Lava from a Volcano, very difficult
    to manage.”
                       C.V. Ramamoorthy, UC Berkeley
                 Phases of Execution
                                                                Transaction INPUT
                         Assembly

Planning &                Configuring
                                                                EXECUTION
Preparation               The System



       Web based Services                                          Transaction
                                                                   OUTPUT
                                     RT

• Feedbacks & Reactions:”Slow”
• Example: Periodic updating &
              Maintenance
• Reaction time RT =slow
• Phases in Execution
     • Before Execution: Proactive
     • During Execution
     • After Execution: Reactive

                                C.V. Ramamoorthy, UC Berkeley
        Some Characteristics of Execution
                  Activities
   Sequences: Normal – Exceptional
   Exceptional Sequences can be
       Unexpected & surprising
       Real-Time
       Interactive
       Dangerous/Life Critical
       Non Procedural




                         C.V. Ramamoorthy, UC Berkeley
                        Continued…
   Need
    •   Recognition
    •   Innovation
    •   Improvisation
    •   Ingenuity
   If this is not the case, a machine can execute
    the task




                          C.V. Ramamoorthy, UC Berkeley
      Some Informal models for Service
                 Execution
   D. Rumsfeld’s Rumblings = (Quite Good!)
   Gul Agha’s Actor Model – script based
   Moore & Mealy – Finite State Machine model
    Procedural, Table-Look-up type, Expert System
   Body’s Immune System, Central Nervous System
    Esther Dyson, M.Gladwell (BLINK!),
    Reflex Action
   Ernst Shakelton’s option-Tree SHOT Model
    You pick your execution action based on – multiple
    options. (Endurance, 1914)
   Strategy based Executions
    Confucius, Sun Tsu, Krishna Covey, Drucker, Collins,
    Mintzberg.
                          C.V. Ramamoorthy, UC Berkeley
     Strategy based Executions
                                      EXECUTION
                                      TIME INPUTS

                                               EXECUTION
 INTRNET           STRATEGY                     ENGINE         EXECUTION
WEB BASED            Enigma                                    OUTPUTS
                                               FAST REACTIVE
Application                                       CACHE
  Strategy
 Providers                                     Buffett’s
                                               Remember the
                                               Bottom Line
                                               always

                                               End Goal

    • An   Application of Pervasive Technology


                         C.V. Ramamoorthy, UC Berkeley

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:21
posted:7/5/2011
language:English
pages:53