Abstruse arguments on EA frameworks
Been having a long interaction on Twitter over the last couple of days with Jon H Ayre (otherwise known as ‘The Enterprising Architect’, @EnterprisingA) and others, about enterprise-architecture frameworks.
(If you’re not interested in enterprise-architecture, probably best to skip this one, as it’s gonna be long; otherwise click on the ‘More info…’ link.)
The starting-point was a post by Jon about the next part in his ‘EA Quick-Start Guide‘:
EnterprisingA: Have just posted Part 3 “Getting Down to Business” of my quick start guide to EA focusing on the Business Architecture http://bit.ly/4UKiCM
The key image of the framework (which I emphasise is © Jon H Ayre) is this:
I like Jon’s work a lot, so duly re-Tweeted Jon’s announcement, but added the comment:
<is for IT-centric ‘#entarch’, but still definitely useful
Which kind of triggered off quite a storm of back-and-forth Tweets:
- EnterprisingA: @tetradian It is not IT arch. Doesn’t even mention it! It is business arch successfully used by bus
- tetradian: @EnterprisingA to me, anything that uses the structure ‘Business’, ‘Application’, ‘Infrastructure’ is an IT-centric arch // Real *enterprise* architecture starts several steps beyond org; some activities may not touch IT (‘Solutions’, ‘Data’) at all // You have one of the best descriptions for IT-centric architecture, but it doesn’t go far enough for *enterprise* architecture // Crucially, your frame barely touches the people-space – knowledge & information vs data, manual-processes vs IT-apps, etc
- EnterprisingA: @tetradian Forget you ever heard of IT then read my guide again. E.g. Trucks are infrastructure. Paper ledgers are applications // There are solutions without IT and data without IT. These are english words borrowed by it. // [If] all you expect to see is IT then that is what you’ll see. But i didn’t mention it.
- tetradian: @EnterprisingA yes, agree re truck=infrastructure, but when would you hear of a paper ledger described as an ‘app’? also where are people?? // …so yes, I take your point, but it still seems a bit of a stretch, and nothing like explicit enough. For me, anyway. 🙂 // point is about language used – non-presence does not help, we need to be explicit (apols for being a pain on this, BTW)
- EnterprisingA: The problem is not the words. The problem is that IT think they only apply to IT. When i talk to business they don’t see IT. // If i ask someone to describe their solution and they only tell about IT parts is it my fault or theirs?
- tetradian: given the huge predominance of the (erroneous) belief that EA=IT, we do need to be explicit each time we use the term // ..so yes, it’s their ‘fault’ in failing to grasp that EA > IT, but also our fault if we make it easy for them do it! 🙂
- richardveryard: @EnterprisingA @tetradian All descriptions are incomplete. We tell a story based on what we judge relevant to a particular context.
- EnterprisingA: @tetradian Point taken. I will amend my guide to include regular reminders that my model is not referring to IT but the full scope of ea.
- tetradian: @EnterprisingA Great! – apologies for having been such a pain, the aim was to make a (very) good model even better – hope it’s helped?
- leodesousa: @tetradian @EnterprisingA been following your conversation … good discussion. Highlights the challenges we have as EA’s about defns
- pauljansen: @tetradian @EnterprisingA Great exchange! Reminding us Information (only) is the meaning the receiver gives to a message.
- EnterprisingA: @tetradian It did help. Appreciate the feedback. It’s a challenge getting IT to accept the wider use of some words so clarity will help // The people are in various places in the model, but for starters in the Services and Organisation models (the people are the org) // The Application layer was the only name on the grid I had issues with but couldn’t find a better one (given its pre-IT meaning)
And there it settled down for the night. This morning I duly re-read the article with a lot more care, and also taking care to expand away from any implicit but unintended IT-centrism. But I had to admit it still didn’t work for me.
The problem is that it’s actually very close to Zachman’s very first model, which Zachman himself very quickly realised was too limited: a simple structure with three columns ‘What’, ‘How’, and ‘Where’. These columns are almost identical with Jon’s ‘Knowledge’, ‘Behaviour’ and ‘Structure’ respectively. There’s an even closer map to Archimate, whose columns are ‘Information’, ‘Behaviour’ and ‘Structure’, with the layers ‘Business’, ‘Application’ and ‘Technology’ – just about the only difference being that Archimate is still noticeably IT-centric whereas Jon emphasises that the framework needs to extend IT. Which is fine if all you’re doing is a conventional TOGAF-style architecture, which regards the business as ‘the enterprise’, and makes a whole load of assumptions about events, people, and the all-important ‘why’ for the business itself. So, perhaps unwisely, I said so: which triggered off another even larger flurry of Tweets:
- tetradian: r @EnterprisingA reviewing your model again http://bit.ly/5fVaqu , it does align very well with Archimate http://bit.ly/7wYASP … // …but it provides no space for what I’ve found utterly essential for all #entarch: the question ‘Why?’ // r @EnterprisingA Without a rigorous, insistent emphasis on ‘Why’, we’re likely to build an #entarch that is efficient but useless // r @EnterprisingA The question ‘Why’ must extend far beyond the org, to describe the context in which the business and its #entarch exist
- enectoux: @EnterprisingA @tetradian instead of asking for solution directly. Try to express your need.
- EnterprisingA: @tetradian Archimate aligns with TOGAF & my model to an extent as they all share some common wisdom (with variations) p.s. I like Archimate // The “why” is required to support the influencing and comms activities but is not part of the core model (which gives the answer). // Rigorously and insistently focusing on “why” can happen without it having to end up as part of the resulting model. // Just because there isn’t a box called “why” and a box called “who” doesn’t mean they aren’t embedded in the models.
- richardveryard: @EnterprisingA @tetradian “WHY” can be outside the model only if we assume a single centralized “FOR WHOM”. (viz homogeneous enterprise).
- tetradian: r @EnterprisingA: agree re Archimate, disagree strongly re ‘Why’ – architecture will fail if ‘Why’ is only implicit, esp. at Vision level // If ‘Why’ and Who’ (and ‘When’, and RV’s ‘For Whom’) are not explicit right from the start, where are they in the models?
- tetradian: @richardveryard @EnterprisingA emphasis on ‘For Whom’? – we would then need to ask ‘*why* for whom?’ – ‘why’ needs primacy! 🙂
- EnterprisingA: @tetradian Architecture is one piece of the jigsaw, not whole shooting match. For context read http://bit.ly/186KOb & http://bit.ly/soIrc // @tetradian I feel (perhaps wrongly) that you are assuming the Architecture World Domination position (I am not a fan of this position).
- richardveryard: @EnterprisingA I think you are being unfair to @tetradian . It is plurality of stakeholder and actor that forces these things to be explicit
- tetradian: r @EnterprisingA in http://bit.ly/186KOb you say arch-skeleton must be broad, yet here you’re artificially narrowing it – why?
- EnterprisingA: @tetradian In my Bus layer “Organisation” model shows who, when, where and for what purpose. “Service” model shows what and for whom. // My Infrastructure layer shows with what and how. My app layer shows in what order/sequence and for what purpose.
- tetradian: r @EnterprisingA http://bit.ly/4UKiCM is a great frame for TOGAF/Archimate detail, but does not provide space for Enterprise-level ‘Why’ … // …Enterprise ‘Why’ etc is not ‘ArchWorldDomination’, but skeleton/frame large enough to describe business *context* … // …TOGAF etc fail as EA models where they fail to provide space for enterprise context (3 or more layers beyond ‘Business’) // …all of What/Who/Why etc (see http://bit.ly/NZYS3 ) need to be explicit right out to all Enterprise layers
- EnterprisingA: @tetradian Agree absolutely; we need to ask “why”, but we are discussing where it goes in the model and in what form.
- EnterprisingA: @tetradian Broad yes, answer yes, working’s out/reasoning? Not necessarily – depends on environment and context. // Perhaps an example of what you consider to be “why” is required for clarity? It is possible we are agreeing without realising. // An example would help as I believe I have that covered but might be misunderstanding what you are saying
- tetradian: r @EnterprisingA examples see slidedecks ‘Vision Role Mission Goal’ http://bit.ly/13xBGd and ‘What Is An Enterprise’ http://bit.ly/8wWNSq .. // ..just about *none* of these are covered in your Business-layer descriptions for Services, Organisation or Information // Remember we’re only talking about the skeleton/frame here – reality is that the frame will never be fully populated.. // ..arch-frame *must* cover whole-Enterprise scope, or architecture will fail wherever it makes assumptions across boundary
- EnterprisingA: @tetradian Then either I have not been clear enough in my descriptions or the detail has not yet been documented in my series of articles // For fear of becoming repetetive, and example of your choosing would help, and then I can explain where it fits?
- tetradian: @EnterprisingA prime example: Enterprise vision/values vs org/business vision/values, and interactions between: where is that in model? // next example: anchoring between market, business-model (#bmgen http://bit.ly/OzZh0) and Business-layer services // Lower-level example: distinction b/w function & capability in services (are merged in IT-services, but separate in manual)
- EnterprisingA: @tetradian Once more I agree (as I do with most of what you are saying) but IMO this is covered by my model.
- tetradian: @EnterprisingA *where* is it “covered by my model”??? – at present it’s all implicit, and does not cover enterprise-beyond-org at all
- EnterprisingA: @tetradian Too general for me still (sorry) – a specific example of two such interacting visions?
- tetradian: r @EnterprisingA conflicting visions/values (real from current client): business profit vs clients refusing to be source for that profit! // unify via shared vision of ‘managing risk/opportunity’ (otherwise loss of trust/reputation – real issue for my client) // Need to carry links to that unifying vision right down into lowest-level details of architecture, as stable ‘guiding-star’
- EnterprisingA: @tetradian If I understand your example, covered in linkage between the Service model (in bus layer) and Capability model (in App layer) // I’m still laying out the approach at the moment. There is more depth to come, but I am laying out the broad model at the moment.
- tetradian: r @EnterprisingA if you leave it until the links b/w Service and Capability, you’ve left it far too late – must be in from the very start
- EnterprisingA: @tetradian I agree with your last 5 tweets and I hold that this is covered by my model (and I wouldn’t suggest leaving the links til last)
- tetradian: @EnterprisingA great! (still disagree slightly, but what the heck! 🙂 ) – apols again for pushing so hard, your model is worth the effort
- enectoux: @tetradian Guys. I tried hard, but it’s a bit hard to follow you conversation. Too much ideas & misunderstanding at the same time… // Right.This is basis. Now challenge is to describe it through the right model & translate it to make it understandable by business // Get down to the atom level: business object.
- enectoux: @tetradian I guess you are speaking of goal model. Still unsure to understand you right.
- tetradian: @enectoux apologies for long/too-detailed conversation – there were (are) some fundamental differences there in #entarch philosophy // Agree “get down to the atom level”, but business object applies only at business layer – other layers have other atomic objects // “I guess you are speaking of a goal model” – goal is the lowest layer, but also goes up above e.g. Business Motivation Model
There we’ve left it for now. I’m still far from happy: at least it is a model that understands some of the world beyond IT, but to me it doesn’t go anything like far enough – as I said in the Tweets, there’s no real space for events, for people, or for ‘why’, nor any real means to handle physical objects (as in Archimate and BPMN, we’re sort supposed magically to realise that it’s an analogue of ‘Knowledge’, and we can put it in there). Yet I have to admit that it probably is the level of simplicity that newcomers to the field will need: what worries me is that they’ll get used to that simplicity, and think that that ‘is’ EA, when in reality the picture needs to be a lot bigger, and should really have been that larger scope right from the start.
Enough – this is a long enough post already! But it did at least seem worthwhile keeping of a record of the conversation, anyway.