David Chappell

Opinari

Get the Feed! Subscribe

Windows Workflow Foundation and BPEL  
# Wednesday, February 28, 2007
 
Two things are worth pointing out about Microsoft's recent announcement of BPEL support in Windows Workflow Foundation (WF). First, it's not a surprise. The company has been talking about its intent to do this since WF went public in the fall of 2005. The only real surprise is that it's taking so long.

This delay is probably indicative of the second point, which is that no one should interpret the announcement as an embrace of BPEL-based development by Microsoft. True, WF's BPEL activities will let developers create workflows that can be directly exported as standard BPEL. But the developer sees those workflows in the usual WF way, i.e., as .NET-based code, rather than as XML-based BPEL. Similarly, any imported BPEL workflows will be converted into WF's internal representation. Like BizTalk Server today, WF treats BPEL as a way to move process logic between different workflow engines, not as an executable format (and certainly not as a development language).

If the popularity of BPEL in BizTalk is any indication, we shouldn't expect widespread use of WF's BPEL support. I very rarely run across organizations that are using BPEL with BizTalk Server today, and I remain skeptical about BPEL ever achieving widespread popularity. Adding the ability to export and import BPEL workflows to WF--and thus to Windows itself--will help WF in situations where support for BPEL is a political necessity. Yet I'll be surprised if it becomes a widely used aspect of WF applications.


14 comments :: Post a Comment

 


Comments:

Microsoft are definitely on the BPEL train and actually are better with WF, which can be looked as a superset of BPEL.

In fact what are Microsoft internally doing now is merging the WF and WCF teams and building the next generation framework for business applications, which in standards body will be represented in WS-BPEL 2.0.
 

I guess this depends on what "on the BPEL train" actually means. Microsoft definitely isn't on the same BPEL train as Oracle, for example. Oracle's BPEL Process Manager has developers working directly in BPEL when the graphical tool isn't enough. Rather than viewing BPEL as an import/export format, as Microsoft does, they see it as a language that people should learn and use directly (at least in some cases).

I'd also argue strongly that viewing WF as a superset of BPEL is inaccurate. WF's Basic Activity Library is in many ways functionally similar to BPEL, but you need know nothing at all about BPEL to work with it. BPEL-based products and WF can address some of the same problems, but their approaches are different in significant ways.

And finally, WS-BPEL 2.0 isn't really comparable to the union of WF and WCF. Apart from the functional differences, which are huge, Microsoft's approach depends on CLR-based languages rather than the XML-based BPEL. This is an important distinction, since I can't make myself believe that most developers will ever be happy working in a clumsy XML-based language. Using BPEL purely as a format for exchanging process definitions makes much more sense than trying to create those definitions directly in this difficult language.
 

Given that BizTalk only uses BPEL for import/export of logic, perhaps the lack of popularity of BPEL in the BizTalk world might be related to the fact that the BizTalk BPEL import is so buggy and problematical.
 

This is a good point, Rich--BizTalk's BPEL import/export certainly could be better. Still, I'm skeptical that improving this would greatly increase the use of this feature. BPEL just isn't a widely used technology today.

WF's BPEL support looks to be significantly better than what's in BizTalk today. Given that the next major BizTalk release will use WF, it's reasonable to one day expect better BPEL support in BizTalk.
 

What is "Windows Workflow Foundation" , and how to use it on a web application?
 

What's the latest on BPEL vs WF saga?
is the "Microsoft's approach depends on CLR-based languages rather than the XML-based BPEL" still true?

Semyon Axelrod
 

Nice to hear from you, Semyon--how's everything?

And the sentence you quote remains true (although since WF workflows are expressible in XAML, you could argue that Microsoft is just embracing a different XML-based language). My sense is that Microsoft has become even less interested in BPEL than it was a few years ago.
 

Hi David,
thanks for the quick response.
I am fine. How about you?

>>although since WF workflows are expressible in XAML, you could argue that Microsoft is just embracing a different XML-based language...

Exactly. So, is XAML as big (in MS world in general) as it was planned to be a couple of years ago or has reality proved to be different?
 

I think it's fair to say that WF hasn't really become a mainstream technology for .NET enterprise developers (although plenty of people create WF-based workflows in SharePoint). Given this, the WF aspects of XAML likely aren't as widely used as they might be.

This is a shame, really. WF is a cool technology, one that's useful for a bunch of reasons (as I argued in The Workflow Way). But who knows? I think there's still a reasonable chance that it will get more mainstream over time.
 

Hi David,
Do you think it is possible to find any metrics that can show me what kind of development team’s productivity gain ( person*months on an average Buz process change) I can expect if the same medium size medium complexity buz app is written with vs w/o WF? I have to sell this to my CFO and w/o any numbers it is hard to do. I am sure you understand where I am coming from.

Thanks again,

Semyon
 

I'm afraid that I've never seen any numbers like this.

I would suspect, though, that a lot depends on the developers and on your future plans. If the team doesn't know WF and if they're not technically strong, the project might take more time. And learning WF for a single project that can be done another way might not be worthwhile.
 

If you look at medium size companies' overall architectures then in my experience a WCF-WF based approach looks like a very healthy alternative, if you could make a hypothetical big bang shift.
A problem seems to be achieving the incremental change, and getting technologists, who despite the rhetoric are often fairly conservative, to move towards truly better ways of doing things.

Moving 50% of line of business operations into lines of Cobol, C, Java and C# over the last 30 or 40 years altogether constitutes a huge global mistake. Businesses will only be able to escape their own inflexibility and overcommitment to I.T. by moving to technologies that are readable and workable by anyone with an operational mindset.
 

I do not think there was much of alternative until 10, maybe even 5 years ago. I have built medium size apps around 2001-2006 using workflow technologies. The first two years were very painful, partly because of immature technology.
 

@semyon

I agree. What WF4/AppFabric needs now is excellent envangelism. It is a real shame that this isn't a much more prominent part of the Cloud marketing.

There may be some economisation benefits in Cloud computing, but in my opinion the one really interesting aspect of it is that Cloud computing should encourage emergence of completely new businesses to replace the existing corporations, simply because large capital expenditure is replaced with operation expenditure.

If this happens it would mean a technology-led deconstruction of the exisitng global legacy infrastructure, but only to replaced eventually with a more centralised version of the same rubbish.

What the WF way offers is a methodological change where the Two Cultures of business and I.T. , that which paralyses the world today, are brought closer together or eliminated. That whole endeavour needs to be a part of the Cloud movement, or the Cloud is just another premature leap towards a socio-cratic ideal that would just fail in overcentralisation, imho.
 

Post a Comment


<< Home