Monday, August 31, 2009

EA without Vision is dead

This is response to an interesting post from Richard Veryard of the Next Practice Research Initiative. Two people I have high regard Alex Buterman and Brenda Michelson commented on it, so I figured it was worth reading. Hopefully this post will be worth reading too. Warning, this was written @ 1AM after watching House, so it might be a bit….1AMish (as in like something written at 1 AM not the religious order who practice living simply in a complex world)

The problem is there isn't a good way of doing Enterprise Architecture (or anything structural thinking aid) without having someone with vision and power. EA is a structure designed to hold information about a business and make it accessible as well as the processes, policies, etc. Blah blah blah. As the Burton Group's Report points out, there is value in being results oriented and aligning yourself to the business and learning to communicate. (the Burton's Group Report was the subject of what the post by Richard Veryard which was commented by Alex and Brenda which lead me to post this! So, its kinda germane)

But what good is that?

I am becoming more and more convinced that EA, and most other IT initiatives fail not because they aren't understood or don't have the management backing or they weren't completed enough. No, I believe the reason is the people involved don't have vision. Most people are not innovative nor do they have vision. (Although they do have perspective!) They regurgitate what others have tried without understanding the physics behind why what the mimicked did worked. If you don't get that sentence. Stop reading!

What seems to be missing in our search for silver bullets and hyper-hyped philosophies is the fundamental principle that all of them (the hyped philosophies, methodologies, do-hickies) are just tools. Things! Things to help us go! But if they that wish to go are without a person who understand where to go inside of their gut, the tools are at best ok.

Take a simple example of SOA. We debate endlessly as to the correct form of SOA. how large, how generic, how encompassing should it be? Should processes be part of a service, consumed by a service or both? These questions are irrelevant, yet that is what people's minds are able to grab a hold. They are stuck in metaphor land. Thinking in terms of the metaphor rather than terms of reality. It's like all those episodes of Star Trek where they said "It's like taking a match to stick of dynamite laced with gasoline." The problem isn't the metaphor, it is the people who believe they understand warp theory now because they know what happens when you take a match to a stick of dynamite laced with gasoline.

Enterprise Architecture, in the hands of a virtuoso, is wonderful because order can be established inside of the processes, activities and stuff that a enterprise does. But the goal isn't order and establishment (well it is for GE and Six-Sigma places who want to take variation out of their processes but that is beside the point) or even to create flexibility and agility. To quasi-quote the "Good Book" (a.k.a. Bible (the Christian one, not the SQL one)) If I have the flexibility of ballerina and have not vision, then I am a nimble dancer in the wrong act of the play. If I have agility and can change the direction of my company on the slightest change of the market, but have not vision, my agility is for not because I will sit and spin with no direction like a child on a sit and spin and spinning until they puke! (Not sure how that last quote ended up there!)

All of our programs, architectures, frameworks, etc. bring order in the chaos. But unless you have someone who pushes you into the chaos your stuck with status-quo.

Brenda has been tweeting about "is a larger or smaller consulting firm better for helping a company 'do' EA (or fill in the blank)" I say, it doesn't matter. What does matter is to have a person of vision to lead us not just in the safe trusted areas, but into the place where Angels fear to tread for there is way to … (Add in your favorite hyper-hyped business term here)

Last thing, If you call yourself an Enterprise Architect, you should buy (that way he gets a royalty) and read Stephen Pinker's book "The Stuff of Thought: Language as a Window into Human Nature." It has a great chapter called "The Metaphor Metaphor." After you read that chapter, read "Down the rabbit hole." If you don't like either of those, he has a great chapter on explicative usage.

The last section of one of the chapters (I think it was "down the rabbit hole") he talks about Plato's cave. Where people are chained head and body made to look at the back wall. There is a fire behind them and people are using cut-outs to make shadows on the walls. The people think "This is reality!" for this is all they know. One of the people escape and find out that they are actually in a cave and real reality is much better. He comes back into the cave and tries to get his brothers out so that they can see the real reality. But they refuse believing instead that what he saw was captured in what they saw on the wall. If you call yourself an enterprise architect, (and actually read this nonsense this far) then you need to be that person who chains are off and can see outside the cave. Use the tools that you have in EA and others frameworks to pry your brothers (or sisters) eyes away from the back of the wall.

Tuesday, May 26, 2009

What does an Event Analyst do?

**** I am working on the images for this post. They will be along shortly! ****

Over the past several months, I have had the privilege of working with some of the top minds in event processing working on a glossary of event processing terms.  Our work has centered around the mechanics and "clock-works" of CEP.  This is extremely important work, but my mind keeps moving towards what does the event analyst do? Perhaps its my ADD, but what's a man to do?

The working group has a diagram of an fully developed event scenario.  We have used this diagram to aid in discussion and as a reference.  But again, my mind turns to the question of "how did an event analyst create this in the first place?"  In this post, I hope to work through that and perhaps evangelize some terms.

First off, what is an event analyst?  I see a specialized business analyst.  A person who recognizes the stimuli that sets things in motion.  It's obvious, a business is replete with business and technical events.  Some of these things like customer order or RFID or check paid are easy to bridge the gap between the real world and the emulating virtual computer world.  However, not all things that happen in the real world of business can be directly measured or captured.  How can you capture the event of fraud or a car accident except perhaps by a human entering information into a screen?  That is what an event analyst does: Figure out how to capture events that are not directly capture-able.

For those still reading, I want to illustrate an example of this and perhaps you can help me make it better. Comments are appreciated!  Especially around vocabulary of terms.  If something doesn't make sense or could be improved, please comment.  This is a work in process, so I have humble opinion of it.  The goal is to make it worthwhile even if it changes from my original opinion.

========  Situation  =========
Emily, an event analyst, works for the state transportation and highways department. She is given the task by her boss to figure out how they can increase capturing speeding motorists while decreasing the number State Trooper patrols.  Getting her cup of Jamaican Blue Mountain, she gets to work.

First off, she knows that there is an business event object in the Transportation system called speeding which is triggers a process in the court system which files the complain with the clerk of courts, sends out the legally approved correspondence and creates the case on the court's docket.  Currently, that event object is published when the patrol-person types the traffic ticket info into the patrol system.  In order to accomplish her task, however, Emily can't rely on human observing the speeding violation.   But how can we get a computer to measure that.  As any good analyst would do, she starts to draw a picture. (well, I would, but I am an architect which means picture-drawer, I believe)

*** Draw a picture of the two speeding events in the real and emulated worlds ***

When she considers the "real-world" event of speeding, a couple of things come to mind.  First off, what object is speeding?  The car.  And what does it mean "to speed"?  It means that the object (car) is moving faster than the legal limit for that section of road.   Emily realizes that the car has a velocity  which indicates the speed its traveling but also, thinking back to her calculus classes about integrals, that at each moment in time the car is at a certain position.  Since an event is a state change of an object, in a sense, the car has a stream of state changes in terms of its position. And the Speeding event would change state to speeding or not speeding depending on the velocity passed  legal limit at that location one way or the other.  Emily adds this event stream to her diagram.

*** Add an event stream for the movement to the picture ***

Emily understands that there is a relationship between speeding and movement which is called an attribute relationship, but how does that help her get the speeding event object in the patrol system to be published?  Perhaps she can figure out a way to get capture the movement.

As she is pondering this, she remembers that Homeland Security implemented a high speed traffic camera that was mounted around some critical infrastructure.  It has the ability to take high-speed digital images of 3 lanes of traffic and capture the license plates of cars.  In Homeland Security's case, they used the information to track the coming and goings of cars around the infrastructure, but Emily had an idea.  The raw events objects that were created had a timecode and a highly precise location.  Being the good event analyst, she reused what was available, subscribed to the raw event objects and started watching.  She know that the individual movement event caused the camera to capture the license plate encapsulating it in the raw event object.  

*** Add the camera event object, the causal relationship between the movement and camera. ***

Homeland security had 50 of these camera setup in the metro area.  All of them were setup at intervals along the roads around some critical areas.  By calculating the time difference between the intervals when a car (as identified by the license plate) past a camera, it is easy to calculate the speed by dividing the distance by the time.  This creates an inference relationship between the set of camera used to detect the movement of the car and the real-world speeding event.

Emily wrote some CEP Script that allowed her to calculate the time difference between the raw event objects for a particular car.  With that information, she was able to publish the speeding event object which acted as a stimuli to the rest of the legal process.  Her finished diagram looked as so:

Sunday, May 10, 2009

SOB - Service Oriented Business

I was having lunch with a friend and he asked me: "What is a simple definition for SOA that I can tell people when they ask?"  I thought about that and all the answers I came up centered around "How do we build it."  

When I googled it, I got lots of answers, but they were very technical in nature or used service in the definition:

OasisA paradigm for organizing and utilizing distributed capabilities that may be under the control of different ownership domains. It provides a uniform means to offer, discover, interact with and use capabilities to produce desired effects consistent with measurable preconditions and expectations. citation

IBMService Oriented Architecture (SOA) is a business-centric IT architectural approach that supports integrating your business as linked, repeatable business tasks, or services. 

GartnerAn application topology in which the business logic of the application is organized in modules (services) with clear identity, purpose and programmatic-access interfaces. Services behave as "black boxes": Their internal design is independent of the nature and purpose of the requestor. In SOA, data and business logic are encapsulated in modular business components with documented interfaces. This clarifies design and facilitates incremental development and future extensions. An SOA application can also be integrated with heterogeneous, external legacy and purchased applications more easily than a monolithic, non-SOA application can. Citation

None of these definitions were something I could memorize and regurgitate when a vice-president asked "What's this SOA thing?"

My definition is simple: "SOA is a way of organizing work that maximizes the consumer / provider metaphor"  A friend suggested I replace "metaphor" with "relationship."

As I was thinking about this definition, I was studying agile programming and there was a reference to the type of company: Functional-oriented vs. project oriented were what the book was referring.  My brain said, "What about service-oriented?"

Can we make a service oriented business (SOB)?  Certainly we have thought of that acronym about employees of a business who believes incorrectly that they are service-oriented.  But can we actually make a business that is completely service-oriented in the consumer / provider metaphor?  What would such a business look like?

Obviously at the top are services that are provided to other entities which they (asuumeingly) pay to consume.  Internal to these services, some work is accomplished.  But probably there are also underpinning services as well.  Services that we rely to make our service perform, stocked, etc.  And those services have underpinning services and so forth.

But if we consider SOA as maximizing the consumer / provider relationship, we could consider the underpinning service as its own business.  A business that "sells" its own products/services.  As a network of these arise, we can see the value matriculating from the end-consumer through our network of inter-related services.

What I find interesting about this concept is that it is value based.  A service invests its income in providing its customers what they want.  I know we have hear the "treating the other employees as customers" rhetoric hundreds of times.  But if we truly created a business as a network of services, would that change?

One other interesting thought that I have about this network of services.  It is ripe for outsourcing.  If we divided a company into logical semantically accurate functionality [work] and consider the "service" as an internal provider of that work and then we found out an external party can offer that same service at the higher value / cost matrix.  Wouldn't that inspire better investment into that service internally?

Said a different way.  Couldn't this help us meet our goal of All work that is necessary for a business to operate should be provided by an entity whose core competency is doing that particular work.

Just some things to think about.

Saturday, May 02, 2009

Other types of functionality... where does it go on the cloud

I have been curious of how will the cloud evolve.  We have picked basic functionality to start the cloud.  Queuing, Storage, instruction processing, network are the blocks that the rest of IT is built upon.  As an industry, those blocks have been/are being solidified.  What I would like to know is what is next to be built or perhaps next after that.  As usual, I have an opinion, but it will, also as usual, require a bit of context.  I hope the conclusion will be worth it!

My contention is this: "The cloud" is a mechanism to enable services of commodity functionality  [Which is another name for work] and as such, a well understood taxonomy or organization of these commodity functionality will need to be developed in order for higher level functionality to prosper in the cloud.  

To understand this, understand that functionality can be described as: Core (value adding), Unique (necessary, but no provider we trust), Commodity (necessary and providers we trust)  and Extraneous (not necessary).  Core competencies (functionality), as described by Jack Welch, is work that a company focuses on being the best at to set itself apart from its competitors.  Unique is work that is needed to be done but it doesn't set us apart directly and there aren't providers that we can trust to do the work.  Commodity is work that is needed to be done that doesn't provide differentiation but there are providers who can do the work.  Extraneous are waste in the lean six-sigma sense.

In real-world products, we have seen business outsource accounting, payroll, manufacturing, sales, IT.  Following Jack Welch's advice of focus on core competencies and allow others to offer a service which encapsulates the work that isn't core to your value proposition.  This makes sense. It leads a company to only, for the most part, focus on things that matter to its bottom line.  However, a company will still need to do unique functionality because no one else can.  Our goal should be to convert, as much as advisable in terms of security, functionality from unique to commodity.  Even potentially creating new industries to provide that functionality.  Stated differently all work that is necessary for a business to operate should be provided by an entity whose core competency is doing that particular work.

So applying this to information technology, there are algorithms that are core, unique, commodity and waste.  If we follow the same advice, our efforts should be focused on our IT's core competency.  Work that is unique we should plan how to make it commoditized.  Commodity should be being done by others and waste should be eliminated.  But how do we make this happen?  

With the advent of services and well-understood interfaces, it has become operationally easier to separate out the core from the commodity services.  The cloud starting from the ground up, with storage, database, computing power, networking.  Functionality that all systems share and because they have stable, mature, well-understood interfaces.  This is not really due to the efforts of the cloud community for they stood on the shoulder of giants. Rather it was the efforts of IT Vendors and standards groups that standardized the interface allowing the creation of their products.  Remember these standards allowed the syntactic ambiguity and also, to the extent possible, the semantic ambiguity to disappear.  However, this canonization effort is expensive, time-consuming and potentially a pre-ripe standard may stifled innovation to soon.

High level functionality such as business rules, calculation, decision logic may indeed be capable of being a commodity.  However, trust, the ecosystem of providers, the organization of functionality and the interfaces standards for the functionality are immature.  Thus it traps potential commodity functionality as unique.  Or it requires that business purchases a large package that stands in for the undeveloped ecosystem which can hinder your IT organization from being enabling business differentiation which should be its primary objective.

SAP and other large package vendors will tell you that they are opening up to allow highly customized processes and manipulation of data.  They will even allow you to call a web service from an external (to SAP) provider (probably will be you!) and use the response in your process.  So, in essence SAP (and all the other ones too) has turned into a platform that provides a rich array of functionality and process to accelerate your mapping of your emulated business to your real business.  This platform can become the focal point of your business.  Is this new?  no.  Is this bad? no.  But the ecosystem needs to extend out.  Potentially even replacing parts of the rich array of functionality or perhaps replacing the platform.  A business using a package should ask itself, what is the core competency of our large package software and map that answer to the previous goal: all work that is necessary for a business to operate should be provided by an entity whose core competency is doing that particular work.  No vendor can provide all functionality to all types of parties as part of their core competency.  At least not to the determent of some of the functionality.  So, how do we build an ecosystem that will allow best of breed augmentation using external functionality?  How will we know who has what?  And how difficult will it be to have multiple providers or to switch providers?  Asked another way, how do we abstract out the functionality from the provider?

These, I believe, are the core questions to prompting us to make high order functionality available on the cloud.  We tried UDDI, but there wasn't a good taxonomy developed to describe where the functionality fit.  As well, there wasn't a good semantic ontology developed to describe the interfaces.  Nothing helped us be agnostic to the physical interface.

So my solution:

Imagine in a folks-onomy way a provider builds a version of their functional taxonomy.  This would include the semantics of what the functionality does.  As well, the semantics of the interface.

This world would also includes a more generic taxonomy built with help of linguists, computer scientists that is semantically accurate.  These taxonomies will be built, probably, along industry segments lines.  The company accomplishing this intellectual property, will work with individual providers to help them map their own taxonomy to the more generic taxonomies potentially across industry boundaries.  Where holes appear, the company will modify the generic taxonomies to make it a more complete and living entity.

The beauty of this is when a company wants to consume commodity functionality.  They search the functional taxonomy to find what they need.  Their platform/package provider will jump start their own taxonomy and the individual consumer will modify it to match their particular implementation.  The individual company will a more specialized functionality for its business than its package provider provides.  Because providers publish what functionality they provide and a match can be found at run-time.  Also, because the interface is semantically tagged, the consumer can provide the correct information based on its data's association with the generic taxonomy at run-time and convert the response to its own format.  

Therefore, the physical interface doesn't need to be well-understood, just semantically understood.  Which has been a huge problem all along.  We don't semantically tag our information (as a general rule) to an authoritative source .  In the same vein, we could do the same for describing nouns, processes and events.  It is in organizing our functionality (work) and our information that will allow our trapped unique functionality to become commoditized and therefore meet our goal.  

I think a company that can build and maintain the taxonomy so that semantic accuracy can be a first order principle in software design would be an awesome place to work.  As the cloud grows and encapsulates higher order functionality, semantics will be foremost!  Well, until it itself is commoditized.  

Anyone want to help me build this company?

Monday, April 13, 2009

The Emulated World

The work the glossary group (of which I am a member) of the event processing technical society is incredibly interesting.  How do you take all of the terms of a new, upcoming transformation in the way we design systems and document them in a way as to expand the usefulness and usability of the concept is what we are trying to accomplish.  To do so there are at least a couple of areas: 1) Philosophy, 2) Descriptive/Proscriptive and 3) Mechanics.  (Probably more!)

I want to write a bit about the philosophy behind events and to a certain extent computer science in general.  My contention is that what we are building with computer systems in general is an emulated world that allow us in a query-able fashion have/get/develop an understanding of the real world.

An examples to make this more clear.  

A warehouse has many objects in it.  Each set of objects (Say a pallet) has a location, quantity, item type, an owner and potential endless attributes.  In the real world, you can't stand in the middle of an isle and say "How many widget Xs do we have?" and expect an answer.  You would have to count them.  So, we develop an inventory system that keeps track of actions done to all of the sets of objects: Move, ship, receive, destroy, assemble, disassemble, etc.  The inventory system is emulating reality and allowing the events that occur in the real world to stimulate process in the emulated world that alters the persistence of this information.

So, the problem becomes how do we get the events in the real world into the emulated world?

Lets keep with the previous example.  An early ERP system would have created an order for work to be done (The move, destroy, ship, receive, assemble, disassemble events), some entity would accomplish that action and mark that it was done.  Usually this all happened on a piece of paper called a pick list.  The pick list would be turned in to a inventory clerk who would enter the new information to the inventory system and it would be updated.  This caused problems because of latency of information update, accuracy.  So people got smarter and they barcoded it so that when the forklift operator dropped it off, it would be scanned and updated quicker.  Now we attach RFID transmitters on these items so that they can be tracked with finer granularity and all movements can be even easier to capture.

This is what I call crossing the threshold between the real and emulated worlds.  We have a variety of means of getting information to pass over the threshold.  User or systemic interfaces, sensors and actuators are all means of getting these events across the border.

But what is our goal in all of this?  We want to have as accurate as possible representation of what is happening in the real world in this emulated world.  Why?  Because the emulated is what we can manipulate, model and make sense of what is happening in the very complex real world.

So, what does all of this have to do with CEP?

There are two types of events.  Those we can sense (without intervention) their occurrence in the real world and therefore react-able in the emulate world and those we can't.  The number of events that can be sensed via sensors such as RFID, biometrics, thermometers, red-light cameras are increasing daily.  A lot of research and money are being spent on increasing the ability of the emulated world to "see/sense" into the real world.

But what about those events we can't sense yet?  Traditionally,  we have built user interfaces to allow humans to intervene and "enter" the event into the emulated world.  The problem with this method is the reliance on a person to intervene?  How accurate will that be?  How timely?

Complex event processing allows a different mechanism to determine that event occurred in the real world.  We can compute it.  We derive an event that we can't directly sense through the combination of events that we can sense.  As an example, we can't directly measure that a car is speeding. (Although that is probably coming)  The speed event however has implications to events that capture-able.  The most obvious is the radar gun.  A police officer uses a radar gun, determines the motorist is speeding, writes a ticket, that ticket is entered into the system updating the emulated world.  Another method would be to have high-speed camera at certain intervals that take pictures of motorists license plates noting the car (identified by the license) and the time and the place.  Further cameras would record the same car at a different time and different place.  The math would be easy to determine how long it took to travel from point A to point B and if that would be in violation of the traffic laws.

Regardless if we have the technology to implement this particular scenario isn't important.  But it illustrates how an event is not an island.  More to the point there are potentially many ways of determining if this event occurred.  Some of them would be able to pass through the threshold between the real and emulated worlds allowing computation to determine the realization of the more interesting event.

Next post will be about how do we name events.

Thursday, January 08, 2009

What should a Enterprise Architect look to find in a job?

I was chatting with a fellow SOA expert Tim Vibber also known as the SOA Chief.  We were talking about what a couple of guys who want to start an SOA consulting company should do.  This took the conversation to why SOA fails and to a post about SOA being dead.

After talking about the failure points, I wondered what makes a good/great architect.  Tim stats that we need to develop systems to be more nimble.  I couldn't agree more, but the question is why can't companies do that?  Why can't architects accomplish that simple statement of "make systems more nimble"?

Both Tim and I are looking for new opportunities so what are these attributes?  Do we have control of them?

To explain these attributes, I am going to borrow something from one of my favorite podcasts: Entrepreneurial Thought Leaders Lecture from Stanford University Technology Ventures Program.  Although I have learned so much from all of the various pod casts, one of my favorite is from Tina Seeliq entitled What I wish I knew when I was 20.  Tina described many important attributes such as Every problem is an opportunity, "the harder I work, the luckier I am" and "never miss an opportunity to be fabulous" and many others.

But the one I want to use here is "Find the intersection between your Interests, Skills and the market."  Tina uses this to describe where one should look to find a career.  She shows where one of those attributes are missing that it is less than ideal.

Interest + market but no skill then your a fan
Interest + Skills but no market then that's a hobby
skills + market but no interest then that's a "job"
Interest + skills + Market = career: A place where you can invest yourself with reward.

For an enterprise architect, I want to change this slightly.  I want to find the traits that are needed for an enterprise architect.

Thus far I've come up with Vision, Passion, Power and Need.

Vision + passion + need but no power = impotent and frustration
Vision + passion + power but no need = unneeded/wasteful projects
Passion + power + need but no vision = future integration opportunity or stuck in the way we've been doing it forever
Vision + power + need but no passion = no buy-in

I would love some thoughts on this.  I think this is very critical to finding good architects and empowering them but also how does an architect find a place s/he should be?