Misc Observations on Biztalk

Comments

# William said on November 30, 2005 4:26 PM:

Absolutely spot-on Bill. Down here in NZ we had the same Biztalk marketing hype.

Seemed odd to me that MS would bother promoting a product that they WEREN'T releasing on the day they WERE releasing 2 others. But it made sense when I thought about it... I guess when you've got a captive crowd, enthusiastic about the 'real' reason for being there...why not push?

And seeing that they have spent gazillions building something over the last 5 years and haven't seen a lot of enthusiasm/uptake of it in the marketplace, I guess I can't blame them, but I could have done without it.

# William said on November 30, 2005 5:01 PM:

Mr P - Thanks man! Do you know anyone with a successful install or is it one of those "loading flat files into a db for 50k" things too?

# William said on November 30, 2005 6:02 PM:

I only know of one company in Australia who paid a fortune to get it setup in '04 and now never use it.

I believe it was used for the first 3 months when they were integrating with another company they acquired, to consolidate figures, reporting etc.. (yeah.. the "loading flat files into a db for 50k" scenario). So in that sense I guess it was a successful install.

But they then moved their systems to a common base and now they don't use it.

# William said on November 30, 2005 8:33 PM:

DoubleI,

I agree. The value proposition for BizTalk is abyssmal. It is outstanding technology, but with a high cost of entry, both in licensing and in applications development. ROI is typically lousy given these costs.

Used properly, it is a great implementation for EAI and BPM. I tend to agree with Mr. Peabody, but maybe in different words. It is easy to measure a value proposition for an initial installation of BizTalk (and ignore the horrible ROI). Upon further usage though, the ability to leverage all of the value of BizTalk is usually missed, especially with ALL OF THE PRETENDER BIZTALK EXPERTS OUT THERE (and no I'm not pointing at you or Mr. Peabody). I can't tell you how many BizTalk sites in Milwaukee (there are only a few) with BizTalk "experts" that haven't ever heard the term "contract-first" or understand the underlying pi-calculus. "Experts", my butt. Most of the sites in Milwaukee still can't get well-formed XML messages with appropriate XSD schema organized for their most basic of applications. They have BizTalk on their resume and therefore are "default experts".

So I think that in addition to the lack of marketing and exposure that BizTalk has received in the past and the lack of real talent in the marketplace, especially with the boneheads out there, BizTalk has suffered. Aaron Skonnard and others of high-profile/highly respected developers/trainers have embraced BizTalk and I think that over time, there will be a maturation in the skill set of developers willing to invest the time to understand BizTalk.

I have been a BizTalker since BizTalk 2000 Beta. I use (extensively) BizTalk 2004 and have converted over all apps from BizTalk 2000. When BizTalk 2002 arrived 3 years ago, there were some, but not many improvements, and therefore not enough to warrant a conversion from 2000 to 2002. 2004 is outstanding and 2006 appears to be even better.

The WCF/WF pairing in WinFX will implement a lot of lightweight features of the current BizTalk offering, and as time goes by, the WinFX offering will strengthen and likely even overtake BizTalk in usage. This will happen in spite of the strong mathematical underpinnings of BizTalk and because of the economic pressures of a free implementation of WCF/WF in the OS. Currently, WCF does not implement "natively" the message bus/broker (categorized depending on whom you ask) of BizTalk, but is capable of the functions/features of BizTalk.

BizTalk, in particular the BPEL stuff, is based on pi-calculus, and therefore has a strong mathematical foundation. Much in the way that relational databases are based on relational calculus and the appropriate transform into relational algebra, pi-calculus is a process calculus that provides, in it's canonical form, 14 FSMs that perform many useful functions. BizTalk implements most of pi-calculus. Google pi-calculus for good info.

The Business Rules Engine in BizTalk is also well-founded mathematically, but I've already wasted a bunch of your comment space.

I'll hope to have some coolness on my blog about BizTalk in December, but I really don't want to speak much past that and jinx the project.

---O

# William said on December 1, 2005 12:12 AM:

(googling on "pi-calculus" now. Interesting.)

I think there are two reasons Biztalk hasn't taken off. One is because it's being marketing, badly, as a shortcut system. Then spend a lot of time on the "don't have to write any code" idea, which anyone who writes programs for a living knows is BS. As soon as you deviate from the ideal path you have to write a LOT of code.

Second, the demos are all of the "take this flat file, convert it to an EDI or HL07 doc, and do the insert". Any first year CS student will be able to do that by the end of the semester. The people doing the demos give the impression that they understand they product and how business processes work, but I never get the impression that they really grok how hard it is to model them.

Plus, they've been hyping it for so long (I first heard about the product at PDC 2000) and not delivering the functionality they promised (drag and drop workflow using Visio was in the demo I saw at PDC 2000.) that I don't think anyone really believes in the process anymore. Most people look at it, see the way it is demoed and presented, and say "Isn't that what DTS is for?"

# William said on December 1, 2005 1:11 PM:

OS - you never cease to impress, never. I personally haven't seen the 'experts' out there.. most of the folks I see by and large make jokes about how much trouble they have with Biztalk. With HL7 or EDI, I think it takes a good bit of technical knowledge (G*d knows I went through h3ll learning HL7) so it's really hard to imagine people working with a tool without understanding the underlying format. Your comments about XML and XSD seem to confirm that. Jeez, if you can't get somethign that easy down, they really need to avoid HL7 like the plague.

Also, I feel like a total dumba33 b/c I don't know pi-calculus from my butt hole. Pi calculus is obviously more attractive. I love RDMBS systems b/c of the mathematical basis, so I guess when I figure out what pi-calculus is, I'll like Biztalk a lot more too.

# William said on December 1, 2005 1:12 PM:

Scott- I'm goggling on it too. The mad scientist is always good for an eye opener like that. Anyway, got a quick question for ya.. have you seen any real HL7 demoes with it? If so, who was doing it, I'd *really* like to talk to them about it.

# William said on December 1, 2005 1:13 PM:

Mr P: Yep, sounds like most of the installs i've heard of, guess some things are exactly the same down under ;-)

# William said on December 1, 2005 5:14 PM:

DoubleI,

As I have not seen your butt hole, I cannot confirm in public that I, too know the difference between pi-calculus and your butt hole. So in the "compare Bill's butt hole to pi-calculus" competence test, we share a similar level of competence. I'd also remark that in this competence test, I'd yield and prefer that your competence exceeded mine.

BTW, the reason that I respect you and your integrity is that you are careful when you declare your competence. You don't use words like guru or expert unless you refer to a skill set that is better than 2 sigma (on the right side), especially when you refer to your skill set. You are first rate in many things and I have a great deal of respect for your breadth of knowledge, especially as evidenced on your other .NET sites. I'm (obviously) annoyed by those in my region that don't choose a similar personal policy. If I suck at something, I either get better at that thing or I make certain that people know I suck at that thing.

---O

p.s. I'll try to refrain from using "suck" and "Bill's butt hole" in the same comment in the future. Actually, I just read your post, "I'm not making this up, honestly" and I think that I would use those words in reference to the letter's author.

# William said on December 1, 2005 6:42 PM:

Never seen a real HL7 demo done with Biztalk. They always mention it. All I've seen are the accelerator product pages at ms.com.

http://www.microsoft.com/biztalk/evaluation/hl7/default.mspx

http://www.microsoft.com/biztalk/evaluation/hipaa/default.mspx

I had one friend that studied the hipaa accelerator, and hipaa in general, for a project he worked on but he didn't use Biztalk in the end. I forget why? I think he said it was just as much work and cost just as much for him to create a custom import as it would have to have used Biztalk.

re:pi-calculus - I've worked with decision networks before, fitness landscapes and what not. I had a boss who always used boolean hypercubes in his presentations. I've never thought of those processes, or relational databases for that matter, having a mathematical basis since I don't have an extensive math background. I learned just enough calculus to get through rate kinetics in my Biochem courses and just enough statistics to get through analytical chem. t-tests are my friend. :)

# William said on December 2, 2005 12:07 AM:

Boy have you guys got your heads up your a*ses!

We have built a very sucessful company around BizTalk and its uses. We have delivered over 75 projects using it in the last company year and haven't once moved a flat file to a database.

In the past we have delivered the UK Governement Gateway - a complex interdepartmental governemnt access tool, delivered an anti-fraud tool for Visa and helped Shell Aviation solve a major Billion dolar problem. With Lockheed Martin we are solving Business Process problems and House of Fraser supply chain issues all based on BizTalk Solutions.

And by the way we've done thew 'I can code it quicker and cheaper my saelf' challange and won.

At 50K its cheap when you look at Tibco, Websphere et al.

So guys get those heads out and get on the gravey train. Look at the surveys BizTalk skills are much most profitable than just dotnet or SQL.

On second thoughts stay as you are and we'll come and plunder your business!

The Brits are coming and we are looking for 200 years of revenge!

# William said on December 2, 2005 12:13 AM:

Hey, the guy who said he could do it cheaper and just as fast himself works with Microsoft as a SQL trainer and helps develop their MCDBA exams. (shrug). I know we've got some opportunities to import some docs from a medical record system that supports HL7 and it'd be easier for us, we're a 2 man operation, if Biztalk could help us. We just haven't seen any evidence that it'll be any easier.

If you've got some case studies of projects you worked on, you limey prick, post them so I can read them. Otherwise take your bad teeth and bland food back across the pond. ;)

# William said on December 2, 2005 2:32 AM:

@SolidSoft Andy,

You said:

". . . Boy have you guys got your heads up your a*ses! . . ."

If it wasn't clear, I'm pro-BizTalk. SolidSoft is the exception. I stated:

". . . and exposure that BizTalk has received in the past and the lack of real talent in the marketplace, especially with the boneheads out there . . ."

That lack of talent comment wasn't directed at you or your firm, but rather at the general market for BizTalk. That there is a lack of developers with the level of talent at your firm is one of the difficulties that BizTalk, as a product faces. A further difficulty is the existence of incompetent developers that represent themselves as competent or even guru level developers in BizTalk. They tend to win projects on resume entries rather than on competence.

You wrote:

". . .Look at the surveys BizTalk skills are much most profitable than just dotnet or SQL . . ."

Within the last week, I was told by the Lead Enterprise Architect of a major insurance firm (for which I consult in many areas) that they were unwilling to pay a premium billing rate for BizTalk work (in any form). They have approximately 20 BizTalk applications in production. This is a typical perspective in this town. They consider BizTalk development, while scarce in developer competence (of which they are fully aware), a commodity skill in their valuation schedule.

Knowing that SolidSoft has exposure in capital markets, I'll assume that you've seen derivatives modeling. While derivatives are traded in double-auction markets, the seller (writer) is the final arbiter of price. If a seller feels that they are not receiving the compensation for the risk incurred by the derivative contract, they refuse the deal. In the same manner, one would think, as you suggest, that BizTalk developers would have pricing power in this market (as sellers of the product, BizTalk Skillset). I'm sorry to say, that they do not. I do not know if this is the mean, but I live in the 12th largest city by population in the U.S. and this condition is likely representative of the state of affairs (and no, don't think of questioning my statistics background; I recognize that there are not enough degrees of freedom to make this assertion).

I do honestly believe that within a 3-5 year window, that WinFX will have provided, albeit in lightweight form through WCF and WF, a platform that will compete with BizTalk as part of the core "included" operating system. The economic pressures of that "free inclusion" will likely prevail, over any likely technological advantage. I don't consider this a losing outcome for anyone, including Microsoft, developers, clients, etc.

And please, don't engage me in debate over the merits of BizTalk over the lack of message bus/broker/bridge/gateway/router (pick your EIPattern) in WCF and WF. That is only a current view of the state of affairs.

I completely agree with your assessment that BizTalk when put in direct comparison with TIBCO, See-Beyond, WebSphere, etc. is the best product and provides the best value proposition and ROI (did you say all of that?)

I don't think that the issue is WHETHER BizTalk represents a viable development platform, but is a question of the PERCEPTION that the BizTalk platform is viable.

All of that being said, plunder away. I SINCERELY (and I do mean it) welcome your BizTalk competence in my neighborhood. It would be a breath of fresh air to interact and collaborate with those that have a bit of depth in BizTalk.

---O

# William said on December 2, 2005 6:59 AM:

Andy J: Ok, you answered my question. You're one of the only people that have been able to come up with concrete things that they've actually done with it. I and I don't think anyone else was trashing it, at least not intentionally. Would you agree though that if it's going to be marketed like it is, that there really needs to be more compelling demos and more prominence of folks that have successfully used it to save money or do things they otherwise couldn't?