When we talk to the business, should we sell SOA?

MSDN Blog Postings » When we talk to the business, should we sell SOA?

MSDN Blog Postings » When we talk to the business, should we sell SOA? — Tue, 14 Aug 2007 10:17:33 GMT

PingBack from http://msdnrss.thecoderblogs.com/2007/08/14/when-we-talk-to-the-business-should-we-sell-soa/

re: When we talk to the business, should we sell SOA?

Ryan CrawCour — Tue, 14 Aug 2007 13:46:37 GMT

Well said ... all too often people sell SOA as the next silver bullet hoping to solve every single problem with it. This is not the case and if people understood that SOA is nothing more than a good solid logical flexible design I think it would receive far greater acceptance.

Some people already have SOA, they might just not know it yet.

I fear SOA and ESB etc. are being branded with the same blood stained label as CRM, the other IT buzzword that has delivered nothing but high project costs and disappointments.

re: When we talk to the business, should we sell SOA?

Loraine Lawson — Tue, 14 Aug 2007 16:54:12 GMT

I disagree with the representation that this is my argument, particularly assertion A. My assertion would be more correctly stated as:

Assertion A. SOA will involved additional IT spending. This could be retraining. It could be middleware - yeah, I know, but it could be.

Assertion B: You will be talking about SOA to the business. (Proofs: You may be talking about money or project delays due to the change in development process. You may also be talking about SOA because, according to some statistics I've seen, the business is interested in SOA and may ask about it. You will also have vendors selling so-called SOA-based solutions.)

Assertion C: The business will have questions. Maybe not line-by-line questions, but if they are asked to spend more money or told about this change at all - see Assertion B - they will have questions.

Assertion D: IT will have to answer those questions, due to the hierarchical nature of business.

My Conclusion: It's in IT's best interest to be prepared to talk about the technical aspects of SOA.

Assertion B : All expenses on a project must be shown to the business, who is paying for the project

But after reading this post, I think I missed your broader point about how this isn't a solution to sell, per se, but a internal change in how IT does it's business. I agree - it should be that. However, it's shaking out a bit differently and while the IT architect may not have to have these conversations, I think the CIO will.

re: When we talk to the business, should we sell SOA?

Edward Hammerbeck — Tue, 14 Aug 2007 23:01:25 GMT

I agree that in general the business doesn't care about how a solution is put together, and therefore, that there is no need to sell SOA. However, SOA adds an some degree risk that does not exist in a two or three tiered, client-server solution wherein everything runs on machines owned and operated by the business, down the hall in the computer room if you will.

Where I work, a very small company with a very small IT shop, I can slap together an application that satisfies everybody's requirements and where everything is all in one EXE. The customer is satisfied, and the project is over. Admittedly, the solution has none of the virtues SOA promises like scalability. So what happens if I, say, abstract the business layer into web services?

First, assuming the web services are hosted internally, I have to have the sysadmins create a space on a web server. Hopefully this web server will not go down or be renamed. Hopefully we won't have DNS issues. Hopefully somebody won't kick out the plug.

Second, assuming the web services are hosted externally or are provided by a third party, how do we negotiate firewalls? What happens if our host goes out of business?

Third, what about security? That has to be carefully considered.

I could go on, but I think my point is made that services, while they have many benefits, they come with risks. A good project manager would have to account for them in the risk planning of any project that used them. The business will have to be convinced that the risk is worth it, and while I would hesitate to use the word "salesmanship" [or some derivative], the project team would have to have a compelling case for chosing SOA over all the other possible architectures that may have other, *perhaps* less serious, risks.

Thanks for the interesting discussion.

re: When we talk to the business, should we sell SOA?

NickMalik — Wed, 15 Aug 2007 07:17:51 GMT

@Loraine,

I hope you are not offended. I guessed at your argument from statements in your blog. If I guessed incorrectly, well... it was a guess! That said, you did pretty much reiterate most of the argument I made in your comment, so I wan't too far off!

--- N

re: When we talk to the business, should we sell SOA?

NickMalik — Wed, 15 Aug 2007 07:29:10 GMT

@Edward,

If I were using SOA to access external services, then I'd be saving a boatload of money! I think I can afford the time to consider security... it is still an IT consideration.

As far as putting SOA on a web server... if you can solve the needs using a simple EXE that you don't have to maintain, for one user to use, go for it. Do the minimum to meet the business need. (Agile).

On the other hand, if the minimum needed to meet the business need REQUIRES scalability and agility, or is part of an enterprise solution that could provide these elements to another app, finding room on a web server is the least of your worries.

If you decide to abstract the business layer into web services, you have created a client server app. Welcome to 1992. All the issues (server goes down, plug not pulled) are client server problems. SOA does not add that risk. These are red herrings.

SOA is net neutral on cost. Loraine assumes that retraining is an expense, but honestly, I've never spent a dime on retraining a dev team on SOA. Although I do think I will spend some money this year to cover my travel... one person in a 6,000 person IT org, four years in to a SOA rollout. That's what... 25 cents per developer per year? Compared to a potential savings of 10-15M this year alone. Gosh... how can anyone afford that?

I live this every day. I don't need to sell SOA to anyone on the business side. I DO need to create a consistent SOA message for IT. In a large place like Microsoft, that's a full time job... mine.

re: When we talk to the business, should we sell SOA?

Mike Kavis — Wed, 15 Aug 2007 07:47:55 GMT

Here's my opinion on this much debated topic.

http://blogs.ittoolbox.com/eai/madgreek/archives/selling-soa-a-true-story-18349

SOA is not a disruptive technology (selling SOA, part three)

Inside Architecture — Wed, 15 Aug 2007 17:43:40 GMT

I've been having a really fun discussion these past few days with some folks on how (or what) to sell

re: When we talk to the business, should we sell SOA?

Francis Carden — Thu, 16 Aug 2007 18:52:17 GMT

Lately, there seems to be more census of agreement on SOA when it comes to business and IT then I've seen in a long time. Thank the Lord.

SOA is a re-architechture, plain and simple. No different a set of problems then if you'd embarked on this last year or 20 years ago.

Reality is, there's a bucket load (100's of thousands) of older applications that are NOT going away any time soon. SOA does not fix that this for most, this side of 2015... SOA for the RICH - see my blog... Great debates.

re: When we talk to the business, should we sell SOA?

Darrell M. Legault — Thu, 16 Aug 2007 19:20:40 GMT

I enjoy your thoughts SOA and how it’s related to the business side of things. As you alluded to, it is a level of trust from the business that they are dealing with professionals, and that “we” know how to solve their needs (much like we trust the accounting department). It is as simple as sell the solution.

I would like to add one more dimension to your thought. From the outside looking in (consulting world), many times you not only sell to the business, but IT as well. One underlying assumption you made was that IT believes that SOA is the right way to go. Not everyone is there yet. So that said, I think the real key is knowing when to sell SOA. Many IT folks do need to be sold on SOA. The existence of FUD (fear, uncertainty, doubt) forces many technical folks down the wrong path. They have a system that is working, but fear to look for improvements, or just do not believe they exist.

I also believe that “What SOA gives you is payoff on the next project”. I need to follow that up with I do believe that for the first time implementers, there is an upfront cost (SOA learning curve). But that is no different than the first time we created our client server app, or our n-tiered app. The cost does disappear on the next project; the difference is those did not provide the payoff on the next project.

There is not a lot of help for them (the non-SOA believers) today. The fact that “SOA is not a product or a package or a system in a box” scares people to death (alternative: believing that your IT providers can do it for you). What if there was such a thing? www.gridgistics.net believes it does exist. Now that sure does change the landscape. Something tangible that provides “good design that makes your apps (and ultimately your infrastructure) less expensive to own”.

re: When we talk to the business, should we sell SOA?

NickMalik — Fri, 17 Aug 2007 01:19:15 GMT

@Darrell,

I completely agree... when you are speaking to IT, you will have to discuss SOA and, if the person you are speaking with does not understand or support the notion, then some selling is involved. From your (consulting) point of view, this is an important element.

Our own consulting services faces this problem daily. When selling to the development manager, we are good at selling tools. When selling to the CIO, we can talk about solutions. In the middle of IT management, and at the developer/tester level, we do need to have a compelling reason why the customer would consider SOA. That said, we should 'sell SOA to IT' only when SOA is the right answer for IT. That is not always the case. SOA is not a silver bullet.

If SOA is the right answer for the client's IT department, and they have no one in the department who knows SOA, then I can see, from a consulting POV, that you would want to account for training expense. Recognize that I'm writing from the viewpoint of an IT department speaking to my own business.

That said, the fact that SOA is not a product or package remains a fact, regardless of whether it scares someone or not. Client Server is not a box or package either. That a vendor, any vendor, wants to sell "SOA-in-a-box" is simply typical.

Don't fall for it. I've seen literally hundreds of those presentations. I have not heard of the company you mention, so I cannot comment on their message. The presentations that I have seen have all been nonsense motivated by the desire to seperate the fearful, the ignorant and the foolish from their hard-earned cash.

--- Nick