The IT Skeptic - Comments for "repeatable process"

HBR and Process

Visitor — Thu, 12 Mar 2009 18:04:00 +0000

The popular press seems to be picking up on this move away from the century old dogma toward rigid process standardization.

The latest Harvard Business Review contains an article which acknowledges that process standardization can, in many cases outside of manufacturing, undermine the very performance it’s meant to optimize.

They call out “professional work”, for example, where the common thread is variability in the process, its inputs and its outputs. The article is a bit basic (and contains a few conceptual errors) but worth a read for those facing process challenges.

http://hbr.harvardbusiness.org/2009/03/when-should-a-process-be-art-not-...

In the long term and at high levels of maturity

JamesFinister — Wed, 17 Dec 2008 21:18:58 +0000

Charles,

In a way yes. When I was a Service Assurance Manager I bridged the gap between architects and change, sitting on the Architecture Review Board, carrying out reviews throughout the V life cycle, and then making authoritative recommendations to the change manager. The SAM role was eventually absorbed into the architectural function when I left, but at the cost of less focus on NFRs.

Architecture is change management

Charles T. Betz — Wed, 17 Dec 2008 14:00:56 +0000

If one has fewer platforms to assess, then changes become less risky and hopefully quicker.

I think that one of architecture's most important responsibilities actually *is* change management writ large - adds, changes, and removals of core computing technologies.

Charles T. Betz
http://www.erp4it.com

Architecture in practice

JamesFinister — Wed, 17 Dec 2008 07:11:34 +0000

If organisation A has an effective architectural function, but poor change management how will their performance compare against organisation B that has an ineffectual architectural function but good change management? Is it even possible to have an effective architectural function without first establishing a strong change management capability?

I think you are right in that at some point the architectural practice becomes the constraint, but I don't think it is in most organisations

Can we move to a model where the capability/capacity of change management drives the resources dedicated to design build and test and the volume of changes in the system? I had one client whose KPI was the backlog of value adding and regulatory changes in the change queue. And in the long term might we want to move to a state where we can pretty well do change on demand?

Architecture, in theory

Charles T. Betz — Tue, 16 Dec 2008 16:21:28 +0000

In theory, having an architecture practice should trim some cycles there, by limiting the number of platforms needing assessment. That's if the architects are focused on value and not off playing with toys. A big if... But this might be a way to measure the value of architecture, in a high maturity shop... interesting...

Charles T. Betz
http://www.erp4it.com

Constraint

JamesFinister — Tue, 16 Dec 2008 10:24:29 +0000

I've worked on the basis in the past that the biggest constraint on most IT shops is the change/service transition process, and within that the biggest bottlenecks are around security and risk assessment because too often they are done just before implementation is due.

Provisioning

Charles T. Betz — Sat, 13 Dec 2008 15:08:19 +0000

War stories have to stay in the bar. But this post was not idle speculation:

http://www.erp4it.com/erp4it/2008/11/mrp-for-it.html

Charles T. Betz
http://www.erp4it.com

In reality?

JamesFinister — Sat, 13 Dec 2008 09:40:34 +0000

Charles,

In reality I suspect they haven't. at least not on any meaningful scale.

I've applied TOC to a few work flow scenarios, but without the client being aware that is what I was doing because it would have been to difficult to explain and get buy in. My attempts to use it in planning ITIL implementations have always been thwarted by in house project managers wanting to use MS Project to do everything.

Go to any ITSM conference in the UK at the moment and you get the impression that every single on going transformation is using Lean, but I'm not sure what is done differently as a result.

Use of maturity models is always talked about at the start and end of a project, but again I'm not sure to what extent using them has led to different decisions than if they had not been used, or a more successful implementation. I would qualify this by saying use of the maturity models is a good way of giving something to make the transformation appear tangible. In my dark days I suspect "Move from CMMI level 2 to level 4" is the nearest that most organisations come to having an ITSM driven strategy.

I would love to hear some actual war stories.

James

Untried?

Charles T. Betz — Fri, 12 Dec 2008 21:23:55 +0000

Have they been tried in ITSM? By who?

Charles T. Betz
http://www.erp4it.com

Goldratt

JamesFinister — Fri, 12 Dec 2008 20:21:58 +0000

I've just been re-reading my review copy of Goldratt's new book "The Choice"

Now bearing in mind that TOC is hardly the simplest of approaches ( I have a little test for all the project managers I work with. I ask them if they know about TOC and if they think it is a good approach. 90% say they do and think it is. I then ask them to explain it to me. 45% pass that test. Then I ask them how they have applied it ....I'm stiill waiting for one to pass that test. ) I was rather taken by this statement:

"...the attitude of most people is that the more sophisticated something is, the more respectable it is...you see since complicated solutions never work people tell themselves that they don't know enough"

My question to the panel is this:

Disregarding that the theories we often discuss here are useful and insightful, do any of them make an actual d**n bit of difference in successfully applying ITSM?

"...how do we manage

Visitor — Tue, 09 Dec 2008 23:13:42 +0000

"...how do we manage performance if processes are non repeatable, in the general case?"

There are a number of methods. Since it appears you are searching for a practical approach to bounding the measurement problem, I'll offer a basic but flexible method.

First, some assumptions:
1 - Measurements are a mechanism for creating organizational memory and an aid for answering a variety of questions associated with the enactment of a process.
2 - Measurements should not be metrics-based but goal-based.
3 - Goals and measures should be tailored to the organization. This requires that the organization make its goals explicit.

The method is called the GQM Framework [Basili]. Organizational goals are identified, questions are developed to determine if the goals are being met, and metrics identified to help answer the questions. It is a mechanism for formalizing the tasks of characterization, planning, construction, analysis, learning and feedback. While useful for many situations, it is particularly concerned with improvement issues.

GQM can be applied to all life-cycle products, processes, and resources. The approach was developed by Basili while at the NASA Software Engineering Laboratory (SEL). It was refined during the 1990s and serves as a foundation framework for many measurement initiatives.

The GQM framework helps solve some of the problems associated with empirical data (measurements do not jive with the reality of desired outcomes). "Dynamic GQM" is an learning based approach for process modeling. It integrates individual GQM models from a global perspective. It is a means of integrating real-world empirical results with simulation experiments.

GQM isn't close to perfect. But it is straightforward and should get you thinking down the right path for the advanced methods.

[Basili] "Software modeling and measurement: The goal/question/metric paradigm" 1992

agnostic

JamesFinister — Tue, 09 Dec 2008 22:28:42 +0000

Charles,

I so, so agree with you. My working definition has always been that process is (at least) technology agnostic. I don't care what accounting package you are using, what I care about is that x days after the end of your financial year we can produce a set of accounts that adhere to basic accounting rules. The IT world is so often driven by the"yes but" view of the world "Yes but SAP does it this way" when the business is driven by "how much tax do we have to pay next month?"

My experience is people start having procedural arguments before the process has been agreed

process as state changes on long lived entity

Charles T. Betz — Tue, 09 Dec 2008 21:51:18 +0000

Exactly. The data foundation is why I periodically ask whether we are discussing process or procedure. (Not that there is a clear industry distinction.) What people criticize as "process" - brittle, overly constraining - I tend to see as lower level procedure.

One way I understand/define true process is "the most significant set of state changes on a major enterprise conceptual entity." Those state changes may be driven by any of a number of procedural variants. The mortgage may be signed with a quill pen or a digital certificate, but there is a state change of "signed" that has remained stable for centuries. Is that the sort of stability we fear leads to the monoculture problem?

Charles T. Betz
http://www.erp4it.com

local and global optimisation

JamesFinister — Tue, 09 Dec 2008 19:08:43 +0000

I wonder if this is why we often struggle with balancing local and global optimisation. We have specialists who (claim to) know a lot about their specific part of the system, but struggle to bring the various point solutions together in harmony with the business strategy. We lack both the data and the tools to make the link between local actions and global results.

Data availability

Charles T. Betz — Tue, 09 Dec 2008 17:35:52 +0000

tangent: "hot topic of lean applied specifically to ITSM."

As I look more and more at applying various management theories to large scale ITSM, the issue of data availability keeps coming up. How can we understand the true dynamics if we don't have the full bill of materials (CMDB with effort & activity tracked by CI) and associated operational telemetry (event management, from raw signals up through incidents, changes, releases, projects, and ultimately the service lifecycle)?

On the shop floor you can walk over to the CAM device and watch it. We have thousands of CAM devices in an all too obscure cloud.

Charles T. Betz
http://www.erp4it.com

it's all about the least restrictive

Charles T. Betz — Tue, 09 Dec 2008 17:20:24 +0000

I don't disagree with what you are saying. Things bite back. I learned about the concept of feedback thirty years ago, while goofing around with loud guitars.

TCP/IP and HTTP still demand absolute compliance within their scope. The innovation was to reduce the scope, to make networking more modular (analogous to microkernel architecture on the OS side). The previous standards were too monolithic, they were proprietary, etc., etc. But "third party innovators" even in New Zealand are still constrained by the global DNS and standards of packet formulation and addressing.

The art is to figure out the minimal number of constraints. Like DNA and the carbon cycle. Libertarians get this.

You still have not answered my fundamental question: how do we manage performance if processes are non repeatable, in the general case? I still think we need lifecycles, value streams, value chains, some kind of invariants at the foundation. Even understanding the things that are changing more slowly relative to others might be a basis. The service lifecycle is a value chain.

Positing a value chain concept does not mean I believe it is fractal, i.e. composed only of smaller linear and rigid chains. Analyzing a repeatable business process does not mean I believe that the entire context it operates in, can also be understood as a process. These may be the bases of our disagreement. Thanks in part to you, I am thinking more and more about the possibilities of systems dynamics to fill in the "white space" problems in Rummler terms. But I am also trying to manage a toolbox for analysis and design. We don't throw out the screwdriver because it is not a drill.

I'm also trying to figure out how industrial theory intersects with system dynamics. For example, I picked up the most recent APICS certification text which talks much about constraints, production scheduling, etc. I expected some mention of Forrester but nada. What is the intersection?

On another cross disciplinary topic, here is an interesting paper on integrating Systems Dynamics with ORM. I'm sure someone will try something similiar with SD & Semantic Web.

-Charlie

Charles T. Betz
http://www.erp4it.com

Again, yes and no. In any

Visitor — Tue, 09 Dec 2008 15:28:32 +0000

Again, yes and no. In any complex system, we always have to be mindful of the unintended consequences (non-linear effects).

For example, biologists have a term called “phenotypic plasticity.” It means an organism can adapt to a changing environment without changing its genetic make-up.

Snails in one pond may be genetically identical to its cousins in another pond. But it has developed a thicker shell because of the presence of crabs. So one is faced with the question of whether identical expressions of a capability may lead to the emergence of a new type of beast altogether.

The analog in ITIL is "substitution." When the creators of ITIL launched their work twenty years ago, I'm sure they did not foresee outsourcing. But their work has become an enabler of it, in particular the many variations of multi-vendor sourcing. New beasts (organizational diversity) are an unintentional consequence.

But to your point, the intentional removal (or perceived removal) of diversity and complexity is ultimately self-defeating. Even when the evidence is staring you in the face, the point is easy to miss.

For example, Charlie Betz earlier wrote, "We build complexity upon foundational constraints. TCP/IP and HTTP demand total compliance, yet look at the blossoming of the Internet."

This is dead wrong and misses the point. There were many prolific network standards before TCP/IP. Remember SNA and DECNet? They stifled diversity and removed complexity by demanding total compliance. Where are they now?

TCP/IP differed because of one critical decision made by Kahn and Cerf. They intentionally split TCP and IP into two separate layers. Why? So the Internet would allow reliable and unreliable protocol services (TCP and UDP).

They called this simple construct the "End-to-End" principle. This network design philosophy allows third-party innovators to express and develop their ideas without the constraints of authorization, regulation or proprietary technologies. This “stupid” network treats all packets as equals. As a result, much of the innovation created in the past decade has come not from corporate R&D or the American government but from young innovators from around the globe--including New Zealand.

Look at the phone company's network of the 1980s and 1990s. They stifled variety. You had to ask permission to use their networks and deploy an application. We saw little innovation as a result.

This ITSkeptic website doesn't exist because TCP/IP is a foundational constraint. It exists because it removed them. It removed the most important one of all: you don't have to ask permission.

There was!

JamesFinister — Tue, 09 Dec 2008 07:09:23 +0000

I guess there are two related streams of interest to me. One continues to be non-process driven approaches to ITSM, and the other is the hot topic of lean applied specifically to ITSM.

I'm sure Madachy's book is excellent, but I don't have the time to read 601 pages whilst doing the day job.

Charles mentioned Rickett's "Reaching the Goal" - I found that interesting, but rather dense for a general readership. Going off on a tangent I'm halfway through Goldratt's latest "The Choice" which I think should be essential reading for management consultants.

My point exactly

skeptic — Tue, 09 Dec 2008 02:12:46 +0000

My point exactly

The more we standardise process and punish those who have their own variants (which may be less efficient in the current conditions), the more we remove adaptability from the system

Yes and no. I think it was

Visitor — Tue, 09 Dec 2008 01:41:14 +0000

Yes and no.

I think it was the movie Jurassic Park where the complexity mathematician remarked, "Life will find a way." Said differently, higher complexity provides greater survivability. Let me explain:

Imagine your kitchen table has 20 legs. Call that a complex system. If we hurl a high-speed bowling bowl under the table and knock out 10 legs, the table remains standing. If the table had only 3 legs (simple system) and we knocked out 1 leg, then the table falls over.

Monoculture crops are like the 3-legged table. If one component of the system fails then the table falls. Take the Irish potato famine where plantings consisted of just one type of potato. When a disruption hit (disease) all the fields collapsed. The famine might have been avoided with a system that treated diversity as a virtue rather than a defect.

However, the conditions for this Irish Potato Famine was created by social and economic forces. A host of economic, demographic, and social pressures marked the decades leading to the famine and meant that the Irish peasantry had no food options when the potato crop failed. Without those conditions, the famine would have simply meant more low-carb dinners.

In other words, they had to work really long and really hard to remove the complexity from the system.

Large adoption rates for ITIL, or any framework or best practice for that matter, serve to create a monoculture of sorts. However, organizations are highly complex; it is in their essence. You've got to work really long and really hard to remove it in such a manner that the organization cannot adapt.

This is what makes change programs so risky. The harder you push a complex system, the harder it pushes back.

The IT Skeptic - Comments for "repeatable process"

HBR and Process

In the long term and at high levels of maturity

Architecture *is* change management

Architecture in practice

Architecture, in theory

Constraint

Provisioning

In reality?

Untried?

Goldratt

"...how do we manage

agnostic

process as state changes on long lived entity

local and global optimisation

Data availability

it's all about the least restrictive

Again, yes and no. In any

There was!

My point exactly

Yes and no. I think it was

Architecture is change management