--- In service-orientated-
[EMAIL PROTECTED], "jeffrschneider" <[EMAIL PROTECTED]> 
wrote:
>
> > Here's an interesting article. discuss:
> > 
> 
http://cio.com/article/438413/Top_Reasons_Why_People_are_Making_SOA_Fa
> il?page=1
> 
> 
> Miko - what a thought provoking article! Thanks for the softball. 
> 
> If these are the 10 obstacles to implementing SOA... we have to ask 
> the very simple question: Can we remove them?
> 
> Here are the 10 reasons, followed up with the 'obvious question'...
> (you may want to read the article first to get the context).
> 
> "Top Reasons Why People are Making SOA Fail"
> 1.    They fail to explain SOA's business value
>  - What if... SOA didn't require an explanation to the business?

It really shouldn't require an explanation, should it. SO applied to 
business architecture may require a bit of an adjustment in thinking 
but by many accounts, SO applied to BA is a natural fit that is 
easily grasped.

Focusing on "the business value of SOA" is incorrect, IMO. The value 
is in defining the BA, SO or otherwise--the focus should be on the 
value of the BA. One may need to justify applying SO to BA, but the 
focus of business value is the BA.

> 2.    They underestimate the impact of organizational change
>  - What if... SOA didn't require organizational change?

The article states: "SOA brings massive amounts of change to an 
organization, especially if the organization does not have a well 
established enterprise architecture in place."

Introducing architectural discipline, in any form, is disruptive. But 
why assume that applying an SO style to EA (I'm assuming a particular 
level here) requires "massive" change?

I agree with the comment by Bert Hooyman in the reader feedback 
section of that article" "...somehow we take it for granted that 
successful introduction of SOA requires organisational change."

Why do we make such an assumption? The definition of the BA/EA may or 
may not identify necessary organizational change. The discussion will 
be in the context of the BA/EA, not simply because SOA is being 
pursued.

> 3.    They fail to secure strong executive sponsorship
>  - What if... SOA didn't require executive sponsorship?

IMO, SOA per se doesn't require any sponsorship because SOA does not 
presume any particular level of abstraction. SO applied at the BA 
level will have the most bang for the buck--but the sponsorship is 
for defining the BA (or EA), not for "SOA."

> 4.    They attempt to do SOA on the cheap
>  - What if... SOA wasn't expensive?

The article says "SOA isn't something you buy, it is something you 
do." And then goes on to list all the things you have to buy. 
Following SO principles isn't necessarily an expensive proposition. 
It is highly likely that an enterprise already has sufficient tools 
(and multiples to choose from) with which to proceed.

> 5.    They lack the required skills to do SOA
>  - What if... SOA didn't require major skills retooling/

The article seems to present SO as though it's a completely brand new 
thing with completely new concepts and approaches--for which we must 
all be completely retrained and consulted, etc. Thinking in terms of 
services isn't the huge conceptual leap that this point seems to 
indicate.

> 6.    They have poor project management
>  - What if... SOA didn't affect project management?

SO doesn't affect project management, IMO. This could be on any "Top 
10 Reasons Why People are Making [pick your favorite topic] Fail."

> 7.    They think of SOA as a project instead of an architecture
>  - What if... SOA was clear in everyones head?

SOA is not a discrete architecture. It is an architectural style.

The author listed "...user interface designers, business process 
models, data services experts, business rules specialists, ESB 
experts, etc." none of which have anything to do with SO. All these 
items are the domain of a robust, cohesive EA. The EA will have 
service producers and consumers as components, and also address the 
other non-SO items.

The phrase "SOA implementation" conjures up images of a project, so 
it might be a good idea to stop using it. "Our SOA implementation 
will be done Q4 of this year." No it won't. First, you're 
implementing a BA/EA, not an SOA (you've got more to deal with than 
just services). Second, implementation of a BA/EA will never be done 
because those items are in constant flux--which is why an SO approach 
can be valuable because it (ostensibly) embraces and enables that 
constant flux.

> 8.    They underestimate the complexity of SOA
>  - What if... SOA wasn't complex?

SO isn't complex. BAs and EAs are.

"It is not a hard concept to understand, but it is hard to implement 
correctly."

Oh if only we had agreement on "correctly."

> 9.    They fail to implement and adhere to SOA Governance
>  - What if... SOA didn't require governance? (AKA, synthetic trust)

SOA governance is too narrowly focused. Not everything is SO related, 
correct? Or does access to the SAP data tables not need to 
be "governed?" ;-)

The focus should be on insuring, through "normal" processes and 
procedures, that the BA/EA/AA principles are adhered to throughout 
the lifecycle.

"...the team must adhere to the architectural guidelines that the 
company adopts."

That says it perfectly. No need for the "SOA" modifier.

> 10.   They let the vendors drive the architecture
>  - What if... SOA didn't need much vendor infrastructure?

Like list item #6, VDA is a universal issue. 

The focus really needs to shift to BA/EA and off of SOA. SOA as a 
term must die. It's a source of endless confusion.

As always, just my opinion. I could be way off base!

-Rob


Reply via email to