Why Soa Does not Deliver

If your IT is supposed to make your business moreproposed in 2000/2001 the consolidation of Inbound
agile, then you need to turn to SOA, right? I do notand Outbound documents into Closed-Loop CRM
agree. Why? Because everyone is once again mixingprocesses with Papyrus WebRepository - to the blank
up cause and effect! An agile organization will fullyeyes of every big-name analyst I spoke to! Today,
utilize IT with or without SOA. SOA will not make ayou find Inbound/Outbound in every ECM vendor's
business or its people more agile.White Paper. Thank you all, because copying is the
Agility will only come from business and IT being willingsincerest form of flattery. Today, I already propose the
to innovate and empower their users by doing awayconsolidation of ECM, BPM, and CRM … So
with rigid business processes and the straightjacket likeyou better stop digging!
impediment of BPM, CRM and ECM. Encoding yourDo we need Governance?
processes into rigid Java for SOA is a business killer.I do agree that we need some oversight to make the
Why the incredible SOA hype?user happy. We can call it governance if you prefer,
The SOA bandwagon is overloaded and subsequentlybut it is simply the requirements of management
bogged because everyone has jumped on to sellbetween users and IT. If you already have a deeply
products and services. No surprise that Networkfragmented IT landscape with too many outsourcers
Computing Magazine has named SOA the mostand external consultants maybe you do need
despised buzzword in November 2006. The 'If Yougovernance to stay afloat.
Can't Sell It, Rename It Game' played by the likes ofSOA or Change Management?
IBM (with Tivoli and WebSphere), Oracle (with Fusion),The core issue of IT is change management. Change
BEA (with Aqualogic) and others. Something has toManagement is about metadata: structured data about
change, and it's not just the product names.data that "describes, explains, locates, or otherwise
Do we even know what is wrong?makes it easier to retrieve, use, or manage an
I seriously doubt it. The influential quantum physicistinformation resource."
David Bohm wrote in 1980: "Fragmentation is soYes, but we do all that with XML, I am told. Listen,
widespread in our society that it interferes with ourWSDL metadata for SOAP and UDDI metadata for
perception and stops us from solving the simplestWSDL is still not enough because neither links to the
problems." Sounds familiar, right? We put everything inuser interface and the process. You need DTDs,
little boxes, buy best-of-breed software solutions forXSLs, XSLTs, XPATH, BPML and BPEL and a lot of
each and every business problem and now we areJava code to verify that the data is valid and use it in
surprised that they won't work together.hard-coded decision blocks.
On the process level we dissect work into littleIn 'SOA For Dummies' the BEA-friendly authors
process pieces because they seem easier to managetherefore recommend a repository for the Java
this way. Process fragmentation was described byprograms AND a registry for the dynamically linked
Thomas Davenport in 1993 as such: "A process is thusSOA services. Their argument for two management
a specific ordering of work activities across time andproducts is understandable in the light of BEA's old
space, with a beginning and an end, and clearly definedTuxedo, the programmed Java Weblogic and the
inputs and outputs: a structure for action. Taking aacquired Aqualogic workflow world.
process approach implies adopting the customer'sHowever, a metadata change in a service interface is
point of view. Processes are the structure by whichNOT automatically propagated to all user interfaces, all
an organization does what is necessary to produceJava modules, all process definitions, all XML
value for its customers."transforms and all databases. There is NO WAY that
I can go along with that but what IT implemented wasa user himself can make any change simply and
more aptly described by Foote and Yoder in 1999,quickly. As there is no common versioning and no
who wrote: "A 'Big Ball of Mud' system is acommon deployment, we are right where we were
haphazardly structured, sprawling, sloppy,before - in a 'Big Ball of Mud' - just more complex.
duct-tape-and-baling-wire, spaghetti-code jungle. TheseStandards - where art thou?
systems show unmistakable signs of unregulatedOthers believe that standards are the solution. There
growth, and repeated, expedient repair. Information isare many SOA definitions from various organizations
shared promiscuously among distant elements of thesuch as the Open Group, Oasis, OMG and others.
system, often to the point where nearly all theMost definitions still miss several core issues, such as
important information becomes global or duplicated."global transactions, security and event handling. As a
This perfectly describes most SOA-Java applicationconsequence there is still no standard way to deliver
servers. Why? Because the clean object-orientedSOA.
encapsulation is destroyed by an irrational requirementSeveral vendors offer SOA Change Management
to use XML messages. XML conflicts with SOA,with tools such as HP-Mercury, IBM Tivoli and others.
because Metadata has to be managed in a repositoryThese are huge investments to deal with SOA
and not inside the data file.infrastructure and networking but not with the
I propose that David Bohm is right and thatfront-end business service needed by the user.
fragmentation is needed for our ability to think, organizeAs you can see, confusion reigns. I see RFPs where
and plan. However, the world around us is not built inSOA compliance is requested at the same time as
process fragments and thus the current SOA'fully-featured APIs'. RFPs request XML and Java
approach will not solve fragmentation but rather,standards, when there is no such thing. That is an easy
create a software engineering nightmare at a higher'YES' response in the RFP. Many IT people seem to
level. Business processes and process changebelieve that SOA in some way solves the
management needs to be first looked at before ITnon-standard situation of Java and XML, when at best
systems can be defragmented.it hides some of the problems.
Business Process 1911's StyleWhere are we going with SOA?
There are those who say that SOA is an extension ofI think that in a few years we will have forgotten it.
process management. BPM proponents are howeverThe big new buzzword is already 'Event-Driven
stuck in 1911's-Taylorism and SOA makes it worseArchitecture'.
because there is still fragmented change management.Why am I not surprised? When I designed
Frederick Taylor believed in fragmentation andWebRepository in 1996 it was built around state
specialization and proposed rigidly structuredevent-driven application models. BPM/SOA vendors
corporations. Hence, each and every IT applicationclaim that you can 'simply connect' events to the 2-D
represents such a business process fragment,workflow graphs of a BPEL process by using Java
because if not, according to Davenport, we don't needcode (i.e. with Oracle jDeveloper).
it. But what is a business process? Rummler andFrom my perspective, to use the word 'agility' for SOA
Brache (1990) proposed that "a business process is athat requires Java programmers to create a simple
series of steps designed to produce a product orevent-driven process is misrepresentation.
service for a customer." And that is quite simply,Consultants and analysts:
incorrect.Why do I seem to continuously needle against
Yes, there are rigid processes and some Ad-Hocconsultants, analysts and outsourcers? I actually do not.
approaches giving more room to the user to chooseBut like everyone else, not all of the ideas that
the course of the process. Strangely enough, hardlyconsultants have are good ones.
anyone seems to realize that the need for an agileService-oriented architecture (SOA) was first
enterprise is created by the dynamics inherent indescribed by Gartner in 1996 in an SSA Research
process related business communication. It is obviouspaper. Gartner, like most research companies have a
that the state of the communication content controlscupboard full of skeletons (failed predictions) and I am
the process and not its meaningless steps. Businessnot holding that against them. SOA is not a wrong idea
communication is not just a document or an email, butand neither is EDA, but what the industry made of it is
can be anything: a selection menu, a web page, awhat I am speaking out against. I am asking you to not
sticker on the document, a data record, images, orlet the prevailing vision cloud your ability to judge for
even a voice recording or video. No matter how muchyourself. I am asking you to not be afraid, to stand up
time and money is spent on business process analysis,and say what you - as someone who is right there at
there will always be one more communication itemthe grass roots of IT - have to say about it. I am
needed for a business process once it gets going. Thisasking users and business department to not interfere
is why collaboration tools and email are now sowith what IT wants to deliver as long as IT has a
pervasive. They don't require analysis to communicate.process focus on the user. Actually, I am asking users
In 'Reengineering the Corporation' Hammer andand IT to be more agile!
Champy made a very important suggestion: "Not theThe Vision of the Anointed:
individual task or process is important but only theWhat is it then that stops most large organizations
outcome." BPM systems and BPR projects miss thefrom forward-thinking strategic change and innovation -
ability to adjust to the dynamics of goal-orientation.as requested by Barbara Gomolski above - equally in
Damelio writes in 'Basics of Process Mapping': "Mapsgovernment or business? Why are otherwise excellent
and flowcharts make work visible ... they represent aprofessionals so afraid to do something new?
snapshot in time …" And that is all that businessI have found one explanation: In "The Vision of the
process is supposed to be, as Bohm said. We needAnointed" Thomas Sowell wrote in 1996 about political
fragmentation to understand, but life does not workvisions in a way that fully applies to other areas
this way.including business and IT visions: "Differing visions, of
How relevant is IT to process?course, are based on differing assumptions …
In 'Does IT matter?' Nicolas G.Carr writes thatFor a prevailing vision, however, meaning that its
'… in the 1990's many strategic investmentsassumptions are taken for granted by the so-called
went to waste … and business executivesthinking-people, those assumptions are never
have grown skeptical about IT, ...' and claims that IT ischallenged with demands for empirical evidence
now a commodity and no longer producing a… A prevailing vision offers more than anything
competitive advantage. I tend to agree, but thea state-of-grace for those who believe in it."
problem is not IT but what the business users requestThe subtitle of Sowell's book is fittingly:
from IT!"Self-Congratulation as a Basis for Social Policy." Only
Users want to organize their work in their ownpoliticians and stock market analysts outdo IT in the art
indescribable way, while management wants to rigidlyof self-congratulation.
control the user process. IT has absolutely no clue howSomeone who opposes a prevailing vision is fought
to go about both, because the underlying technologywith all means possible by claiming that 'the hidden
(mostly because of Windows, Java and XML) hasmotives have to be disclosed.' Much like the dicussion
become nearly unmanageable. SOA simply adds toon climate change ...
the already incredible complexity.References and innovation:
Business users are human and point at something toIT people in large organizations more than others shy
say: This I like and this I don't. Once users see a cuteaway from innovation. I am always asked for
GUI front-end they think the inside must be cute asreference installations (and no one wants to be one)
well. Users buy a GUI, not architecture, flexibility orwhich is impossible for innovative software. Hey, it's
long-term manageability.NEW; no one else has done it before! Additionally
They demand however 99.99% availability (which isthere are no two customers who actually do the
fine for the underlying mainframe-based,same thing with our software or have the same
bank-transaction system); but when the averageinfrastructure. As always, a reference should be about
mental and physical availability of an employee is atthe integrity of the people involved and not about
most 50% then the availability expectation linked withtechnology.
the current complexity of technology creates the longYou can obviously choose the safe route and decide
rollout cycles. Users expect IT to be inhumane andnot to innovate and make your choice from a Gartner
that's exactly what they get. And because they cantMagic Quadrant. Gartner Group is however a market
get what they want business is fighting for controlanalyst company and thus their information is from the
over IT with Outsourcing and Governance. Bad idea!past. You won't find innovation there.
Adherence to ineffective governance rules is usuallySOA is not innovation but an evolution of
more important than what constituents need, in caseobject-oriented messaging that took a wrong turn
you have not yet noticed. On his ZD-Net blog Joebecause vendors needed to sell what they had - they
McKendrick asks to "Turn your SOA into a HOAjust renamed it. My position in this paper has in
(human oriented architecture)" and says, "Withoutdifference translated into ISIS Papyrus products over
governance, there is no SOA. But maybe we don'tthe last twenty years and therefore some of it has
want too much governance." I want to add: Maybe webeen and is once again very innovative.
don't want SOA at all if it requires even more rigidity toI propose that going into a huge programming effort
control and manage our business processes?yourself or buying a rigid piece of 'standard' software
Business Process Decision Making:(with rigid processes) involves a much bigger risk than
Let's face it: The AGILE-SOA-jBPEL-XML-enterprisegiving something new a try. IT is the most powerful
can't see beyond Tayloristic 2-D process graphs! Thecompetitive tool there is - if implemented and used by
latest buzzword of 'IT-Industrialization' describes ouragile people.
plan to create Henry Ford conveyor-belt-IT-processesYou can't lead by walking in someone else's footsteps.
for lobotomized business users.If you don't innovate, your IT is not leading edge no
It gets worse. In 2005 Davenport and Harris furthermatter how buzzword compliant you are.
claimed in a MIT-Sloan article: "Rather than requireBenchmarking IT against others who don't innovate will
people to identify the problems or to initiate theonly pull everyone down to the same low level. But
analysis, companies typically embed decision-makingthen … you can show that you are among the
capabilities in the normal flow of work. Those systemsbest in your benchmark!
then sense online data, apply codified knowledge orInnovation - doing something new - always bears
logic, and make decisions all with minimal amounts ofsome risk. Be brave!
human intervention."Bibliography and References:
In "Super Crunchers: Why Thinking-by-Numbers Is theAllen, Paul (2006). Service Orientation, winning
New Way to Be Smart" Ian Ayres too claims thatstrategies and best practices. ISBN 0521843367.
huge data sets allow previously impossible predictionsAyres, Ian (2007) Super Crunchers: Why
and that statistical methods are more accurate thanThinking-by-Numbers Is the New Way to Be Smart.
the more intuitive conclusions drawn by experts.ISBN 978-0553805406
From my perspective, Business Intelligence producesBohm, David, (1980) Wholeness and Implicate Order,
an unintelligible mess of data that has little to noISBN-13: 978-0710003669
meaning for business decisions, automated or not.Carr, N. G. (2004) Does IT Matter?: IT and the
Imagine that we not only make the processes moreCorrosion of Competitive Advantage ISBN-13:
rigid but also the codified decision-making knowledge?978-1591394440
That will ruin each and every business because theDamelio, R. (1996) Basics of Process Mapping, ISBN-13:
`statistical data facts' are illusions and at best average978-0527763169
values from the past that do not apply to the currentDamasio, Antonio (2005) Descartes' Error: Emotion,
point of decision. Face-to-face interviews with theReason, and the Human Brain, ISBN 0-380-72647-5
relevant managers, employees and customers will giveDavenport, Thomas (1993), Process Innovation:
you an emotional intuitive feel for the right decision thatReengineering work through information technology
statistics won't.Draheim, D. & Weber, G. (2006) Trends in Enterprise
You might ask: Emotions?Application Architecture, ISBN-13: 978-3540327349
I have been asked to replace 'emotions' with 'instinct'Erl, Thomas (2005). Service-Oriented Architecture:
because emotions is seen so negatively. I refuse!Concepts, Technology, and Design. ISBN 0-13-185858-0.
Instinct is more in the area of reflexes which do notHammer, M. ; Champy, J. (1993), Reengineering the
require any thinking at all. Maybe feelings would beCorporation: A Manifesto … ISBN-13:
better, but feelings also require emotions deeper down978-1863735056
in our human nature. I can go with intuition becauseHurwitz; Bloor; Baroudi; Kaufman (2006). Service
intuition is an emotional function.Oriented Architecture for Dummies. ISBN
Today's assumption is unfortunately that emotion or0-470-05435-2.
intuition is the opposite of reason. Maybe that is not so.Johansson, Henry J. et.al. (1993), BPR: BreakPoint
Why can in the UK and the US the anti-depressionStrategies for Market Dominance ISBN-13:
drug Prozac be detected in the rivers? Because978-0471938835
people are continuously forced to be reasonable andNewcomer, Eric; Lomow, Greg (2005). Understanding
not emotional. Look inside yourself and you will feelSOA with Web Services. ISBN 0-321-18086-0.
that it is against human nature and it makes us sick.Pulier, Eric; Hugh Taylor (2005). Understanding
Human decision-making is most likely an emotionallyEnterprise SOA. ISBN 1-932394-59-1
weighted pattern-matching ability of the brain.Rummler & Brache (1990), Improving Performance:
Rationality is only used for post-decisions justification orHow to manage the white space … ISBN-13:
verification. Are not strong managers and978-1555422141
entrepreneurs mostly very emotional, evenSowell, Thomas, (1996) The Vision of the Anointed
unreasonable people? Absolutely! Pioneers such asISBN-13: 978-0465089956
Broca (1878), Papez (1937), and MacLean (1952)Taylor F. W. (1911) The Principles of Scientific
suggested that emotion is related to a group ofManagement, ISBN-13: 978-1434638205
structures in the center of the brain called the limbicWatson, Thomas Jr. (1997) Father, Son and Company,
system, which includes the hypothalamus, cingulateISBN-13: 978-0593020937
cortex, and hippocampi among others. AntonioInternet resources:
Damasio - today Professor of Neuroscience at theSSA Research Note SPA-401-068, 12 April 1996,
University of Southern California - has long researched"'Service Oriented' Architectures, Part 1"
neural systems for memory, language, emotion, andSSA Research Note SPA-401-069, 12 April 1996,
decision-making. In his 1994 book, "Descartes' Error:"'Service Oriented' Architectures, Part 2"
Emotion, Reason and the Human Brain," he documentsSOA Reference Model Technical Committee, OASIS
his discovery that "humans with dysfunctional(2006). A Reference Model for Service Oriented
emotional centers face grave difficulties inArchitecture. (PDF).
decision-making." Thus human decisions - intuitive or notRational Unified Process, IBM, Mar. 2005,
- are emotionally weighted and how do you want towww-3.ibm.com/software/awdtoolsrup.
encode that, Mr. Process-Analysis-Consultant?IEEE Std. 1003.1, Standard for Information Technology-
To improve decision making with the help of computingPortable Operating System Interface (POSIX), IEEE,
we need to invent technology to model intuitive human2004.
decision-making far beyond logical rules. While IanE. Christensen et al., Web Services Description
Ayres mentions neural networks and pattern matchingLanguage (WSDL) 1.1, World Wide Web Consortium
he fails to understand that both do not need statistical(W3C) note, Mar. 2001;
data but decision-points linked to real data. It is notSOAP Version 1.2, World Wide Web Consortium
important how many people took a certain decision,(W3C) recommendation, June 2003;
but what data pattern was used by each individual toSOA Blueprints, Middleware Research, Mar. 2005,
come to the decision. Therefore we do not need toAnderson, A. "IEEE Policy 2004 Workshop, 8 June
crunch meaningless and inaccurate data with unproven2004, Comparing WSPL and WS-Policy," Sun Labs,
mathematical assumptions, but today's computing2004;
power has to be used to interactively learn decisionsBrian Foote, Joseph Yoder (1999)
from humans.Joe Kendricks, (2007)
And what about 'Agility'?Coenen, Alcedo (2006). SOA agility in practice (PDF).
I am serious: Agility is a human property and notVia Nova Architectura.
system functionality. What stops current organizationsJones, Steve (2005). Toward an acceptable definition
from being agile? The Java/XML 'Big Ball of Mud'?of service (PDF). IEEE Software.
Tayloristic BPM? Windows DLL hell? I say it is the lackMittal, Kunal (2006). Requirements process for SOA
of agility in people and their small-minded resistance toprojects, Part 1 of 3: Capturing requirements for an
true innovation.SOA application - Initial requirements to build out your
An agile IT department run by a truly agile CIOSOA (HTML). IBM Developerworks.
supported by an agile board creates an applicationShan, Tony; Hua, Winnie (2006). Solution Architecture
platform that puts great power into the hands of thefor N-Tier Applications (PDF). In Proc. of the 3rd IEEE
skilled business user who is willing to be agile himself.International Conference on Services Computing (SCC
In their SOA White Papers the likes of IBM, BEA,2006), pp. 349-356.
TIBCO and Oracle name an endless array of technicalWada, Hiroshi; Suzuki, Junichi (2006). A Model-Driven
complexities to create SOA when there is not enoughDevelopment Framework for Non-Functional Aspects
IT manpower available on this planet to even do 2 %in Service Oriented Grids (PDF). In Proc. of 2nd IEEE
of such SOA projects (Source: Max Pucher, 2007,International Conference on Autonomic and
informed guess of 35 years IT). Remember, once theAutonomous Systems (ICAS 2006).
hole is so deep that you can't climb out, digging deeperBieber, Guy (2000). Programming Service Oriented
will not improve your situation.Programming (HTML). Motorola.
KISS - Keep it simple, stupid!Gomolski, B. (2006) Computerworld Opinion, Oct.
Simplification is needed, and it was I who first humbly