« SAP TechEd Munich - preamble | Main | Community is Open Source: On SAP Developer and Business Process Communities. »

Comments

Ric

Is a non-repeatable process an appropriate inclusion in a transaction-oriented application? The attraction of an apps package is that it (hopefully) takes away most of the stress around the bread-and-butter aspects of the business - would asking it to cater for ad-hoc implicit-knowledge-based processes be just a bit too square peg:round hole to be effective?

My point of view is that that makes a business too app-centric (and heavily locked-in) - if you want an all-inclusive process model do it outside the app with service-based touchpoints into the app where required for transactions. That way you could change the app (eg SAP to Oracle), or take advantage of advances in the app space itself (R/3 to Netweaver to BBD for instance).

sig

Ric,

agree that the hard part is to find the limit between what needs to be structured and what's better off not being structured - last part has social software, water cooler chit-chat etc.

In fact that question can be split in three: 1. Why focus on transactions, should it not be changes to states of objects? 2. What are objects and thus prone to transactions / events or other object changes? 3. What are events?

1. As you know I'm saying that transaction or event focus limits the architecture of an enterprise system - if you think simple and say that it's all about changing states for objects - in truth adding value to objects - then you're closer to a business strategy. Actually that is precisely what business is all about - add value to objects, tangible or not!

2. If "object driven" then there are no limits to what objects are, all important for your business - widgets, bikes, frames, screws but also issues and solutions if you're in a service industry. Ah, if you have objects like issues and problems then of course anything can be covered, even those pesky ad-hoc things that happens in production.

And in reality, most ad-hoc processes, barely repeatable on the surface, are core process derails and are thus quite important to capture and run.

3. If you stick to "transactions", an event is very precise and limited - typically run and captured by the legacy systems. But who says it shall be like that? There is no reason to limit an event if it's about changing the state of an object. "Go forth around the world and find out as much as possible about xxxx" - an event that could take months, involve many activities but that still could be a single event in a system: The system delivers the object in present state, you go forth and do what's necessary to enhance that object - no always a need to capture all that happens, merely the important stuff and the change to the object.

That would leave room for any method or system to be used within one event delivered by a main system. And with a flexible system; a chance to build all within that event (or part of it) into the main system if in any way repeatable as experience increases.

See, now you got me started on almost a post within a post :D

Ric

Sig - you're either saying that you think SAP is capable of changing their philosophy and can do all that you suggest in your "mini-post", or that they are not, and the world needs thingamy.

My point wasn't that I thought transactions were the best way forward, or even the best description for business events - I am saying that SAP and its ilk DO "transactions" ... and that's all at the moment. Now they do it well, but as you so eloquently point out, that's a long way short of complete from the perspective of the whole business. Increasingly I think (and certainly in innovative businesses) the ad-hoc, non-repeatable processes need to be the CORE, not derailments of the "core" - they should be the norm, not the exception. thingamy can probably handle that level of ambiguity and "randomness" - I doubt that SAP and Oracle in their current forms can

sig

Ric - I think (actually I know..hehe) that SAP is aware at least of the issue, but as you say - they cannot and thus will not.

This due to the architecture of the whole shebang, hailing back about 25 years. Just looking at their business objects is enough to convince me about that.

But as they are aware they do try as well as they can, and I think they may get half there through easier ways to alter the processes (Galaxy seems to enable just that) - but in my humble view it is a hard and complicated issue due to the underlying code and architecture. Where the wall will be nobody knows, but a wall is somewhere there between current and future perfection.

Sometimes one just have to start over, cannibalise one's products. Don't see any signs of that though.

And I think you're right - the not-so-linear processes should be core - suspect that's where the major value adding happens for all firms, while the linear processes could be seen as mere execution of what comes from the not-so-linear processes. Good point indeed!

Thing is that most are so focused on the negatives - "aaargh, my daily problems" - that the not-so-linear processes focused on are the derailment, the daily problems, the noise - stuff that all would like to simplify. Hence perhaps my choice of terms, market-adjusted as they are :D

On an aside, I think SAP are better poised to perfect their product than Oracle as they in my mind have a more holistic view on their product - one underlying platform with all built in and around and not a whole portfolio of loosely connected bits.

The comments to this entry are closed.

My Photo

Contact


  • Phone: +33 6 8887 9944
    Skype: sigurd.rinde
    iChat/AIM: sigrind52

Tweet this


Thingamy sites

  • Main site
  • Concept site

Tittin's blog


Hugh's


Enterprise Irregulars


Faves

Twitter Updates

    follow me on Twitter

    alltop


    • Alltop, all the cool kids (and me)

    Subscribe

    Blog powered by Typepad
    Member since 01/2005