Category: services

(micro-)service oriented architecture

CentraSite portal and blogs now officially launched

I wrote back in July about CentraSite, a standards-based SOA forum and a freely-available product for SOA registry, repository and other functions. This week, I received notice that CentraSite is now launched (I guess that I was playing with beta code before), and that there will be a number of blogs hosted on the site to talk more about SOA. It will be interesting to see how this plays out; it has Fujitsu and Software AG behind it, and there’s obviously been a lot of work put into the product, but I don’t have enough experience with this class of product to tell if it’s a potential winner or not.

IBM’s Virtual Jam on SOA and BPM

The Virtual WebSphere BPM User Group is hosting a 2-day Virtual Jam on the BPM with SOA forum. I heard about this from Bruce, who points to the registration page; note that you have to join the Global WebSphere Users Group Community, then join the Virtual WebSphere BPM User Group (there’s a not-so-obvious link to join on the aforementioned registration page). If you survive all of that, hop on over to the forum site and see what’s being posted; the jam runs through the end of day tomorrow.

Someone from IBM has appeared to have seeded a bunch of discussion topics, but there’s not a lot of participation yet. I’m not sure that a forum is a good place to hold a 2-day jam, since the cycle time of people checking and responding to forum posts can be a bit long for that. That being said, there are a few good topics running.

Bifurcating BPM, Batman!

My reading is a bit behind due to my recent travel, and with my feed reader sitting at just under 3000 unread items, I tend to ignore the magazine feeds in favour of reading blogs. Today, I went back through the unread Intelligent Enterprise items and found The Bifurcation of Business Process Management, which has very little to do with bifurcation except for one sentence in the summary; probably the editor just likes the word. 😉

I don’t agree with everything in this short report from Ventana Research: they state that BPM is still in its infancy, but suggest creating a 3-year architectural blueprint for BPM; if the technology truly is immature, then a 3-year plan would change so quickly as to be obsolesced within a year, wouldn’t it? Also, the phrase in the recommendations, “BPM technology, by failing to maintain pace with Web services technologies, is not ready for prime-time enterprise SOA applications“ is a generalization that I just don’t believe to be true.

I’m left with the impression that the author is not all that knowledgable about BPM, which doesn’t leave a particularly good impression about Ventana.

RedMonk Radio on SOA and mashups

Catching up on some podcasts today while walking around Toronto, and listened to the June 30th RedMonk Radio that included a section on SOA and mashups. Favourite quotes: “Describing SOA as mashups is insulting to mashups” (Steve O’Grady), plus Coté’s analogy “SOA is like religion, and mashups are like science”.

As always, worth listening.

Mashup Camp 2 Day 1: Mashdowns

As I mentioned in my previous post, I had to do all my blogging today offline because of the spotty wifi in the Computer History Museum, and I have to say that Windows Notepad makes a pretty sucky offline blogging tool. However, I’m relaxing back in my room listening to the newly-downloaded and extremely enjoyable Veneer (just available on iTunes, after I couldn’t buy the CD after a month of trying on Amazon.ca), cleaning up the blog posts and paper notes from today.

Following the kickoff session, we headed off to breakout sessions proposed by anyone and everyone during the kickoff. Each session was supposed to update the wiki with notes from anyone at the session, and you can find the grid of sessions here with links to the wiki pages with the notes. I’ll link to the notes for each of the sessions that I attended.

The first one that I headed to was “Mashdowns: mashing for competitive advantage in rich client/enterprise applications”, led by Mike Fisher and Ben Widhelm from ElephantDrive. They see this as a second generation of mashups: more tightly integrated into desktop or enterprise applications, and more focussed on “doing” rather than “consuming” — which seems pretty much aligned with my ideas about BPM and mashups. I hate their term “mashdown”, however, preferring the more-commonly used “enterprise mashup”. Really, the distinction between first and second generation mashups is primarily between consumer mashups and business/enterprise mashups.

We gathered a number of ideas about the difference between first and second generation mashups:

  • First generation mashups are about the “what”, and are primarily about aggregating/joining/federating data. They’re generally seen as useful by users (consumers), and because they’re focussed on the consumer market, they tend to be public, and developed rapidly and a bit loosely. The revenue model is usually based on ad revenues, since few end-users pay for the mashups.
  • Second generation mashups are about the “how”, and are about aggregating external and internal (enterprise) services. They’re useful to business for all the usual business ROI reasons: improving process efficiency, reducing IT costs and increasing business agility; like any other plan that reduces technology capital investment, they also tend to level the playing field for smaller companies since they can use the same technology as the big guys but not have to build it or buy it outright. Unlike the consumer mashups, however, they have to be industrial-strength, private and secure. Equally importantly, they have to be supported by some sort of service level agreement backed by appropriate high availability and disaster recovery scenarios, which most of the current API vendors are not willing to provide.

The key difference for me is that second generation mashups are about integrating into the business processes. This breakout was a significant conversation since it’s the first one that I’ve heard at either Mashup Camp where business processes were a major focus. I’m feeling very positive about BPM and Web 2.0 today.

We had a conversation about one of the main problems of enterprise mashups, which is their current lack of acceptance by IT. Part of this is IT attitudes: not trusting the external APIs, either in terms of data integrity or in terms of reliability, plus the NIH problem. An equally important part is the relatively lack of readiness of the APIs themselves in terms of SLAs, authentication and other indutrial-grade issues that would have external APIs be on equal footing alongside internal ones. Even with internal-only mashups, that use lighter-weight mashup techniques on internal APIs, there’s resistance to a new way of doing things. That really comes back to the question of the the difference between a mashup and any other web services orchestration, especially as lightweight (non-WS-*) integration methods are used for faster application assembly internally.

This was a great session for focussing my thoughts on how to talk to my enterprise customers about mashups.

Michael Scherotter was also there from Mindjet, distributing copies of their application on flashdrives. Haven’t had a chance to install and try it out yet.

Webinar on SOA standards and CentraSite

I’m tuned into an ebizQ webinar on SOA standards, News Break: The First Standards-Based SOA Forum to Manage and Govern Your SOA. This link should be good for replay within a couple of hours after the webinar, or within a couple of days if you want the full version with the live Q&A. By the way, whatever happened to the link that would add the webinar directly to my Outlook calendar? I miss that! I’m simultaneously blogging (obviously), packing for my trip to Mashup Camp and listening to the fire alarms being tested in my building, so this may not be as detailed as usual.

The presenters are Keith Swenson of Fujitsu (who I’ve met a couple of times), Daryl Plummer of Gartner (who I heard speak at their BPM Summit earlier this year), Paul Butterworth of AmberPoint, Peter Kuerpick of Software AG and Jean Francois Abramatic of ILOG. The blurb for the webinar promised:

[A]n exciting multi-vendor announcement on how the leading SOA vendors are partnering to achieve a common SOA infrastructure. This initiative will leverage an open, standards based SOA registry and repository to manage and govern the complete SOA landscape. It will allow you to analyze interdependencies in your SOA including services, processes, applications and other SOA components.

In other words, now that most vendors have figured out that most customers are not going to be using a single-vendor SOA infrastructure, they’re getting together to build some standards in the area of registry and repository.

Plummer started with a message about the importance of SOA governance, and the recent focus on this by many organizations. He stepped through Gartner’s models of an SOA framework and service registry, and touched on policy management and a few other governance-related issues. He laid a lot of the groundwork for the rest of the webinar, since SOA governance is a key driver for standards.

Up next were Kuerpick and Swenson, the two webinar sponsors, to talk about the CentraSite Community, with is both a standards-based SOA forum and a product that provides an open registry and repository, impact analysis tools, and governance tools that store, tracks and analyze processes and their underlying services and interdependencies. They launched a canned demo of CentraSite that is also available on the site, which happens at just below light-speed, so I’ll need another viewing to catch all the details, but it appears to be all browser-based and has some interesting functionality especially around interdependencies of services. CentraSite is already supported by several vendors, and any standards-based vendor should be able to publish directly to it but would need to get a bit more involved to be a full player. It will be interesting to see how this catches on over the coming months, and if it manages to sort out some of the SOA confusion.

One really interesting point is that both XPDL and BPEL are mentioned explicitly, and BPMN was also mentioned although it’s not on the slides and isn’t used in representing business processes within CentraSite as far as I could see in the demo. CentraSite is not a standards organizations, and much of the underlying standards work will be done by the existing standards bodies such as OASIS.

There is a community edition that is free of charge, and you can register to download a product evaluation.

More to come on this in the future, I’m sure.

SOA anti-patterns

This is brilliant: Steve Jones on SOA anti-patterns. I was going to just do my usual del.icio.us link/auto-post, but it’s inspired me to an entire post.

A few months back, I wrote an article for AIIM E-DOC Magazine on BPM Implementation Pitfalls that described my three favourite ways to screw up a BPM project: over-customization, allowing the business to design the solution, and applying the wrong BPM tool. Jones’ article on SOA anti-patterns is more formal treatment along the same lines that addresses the following burning issue:

A lot of emphasis has been placed on implementing Service Oriented Software according to best practices and principles. But how about the worst practices?

He rigourously lays out several SOA anti-patterns that had me laughing out loud, including “The Shiny Nickel” (used to incorporate the latest technology buzz within your SOA for the sake of telling people about it) and “IT2B” (creating “business” services based on the belief that IT understands the business results in services that meet neither IT nor business goals).

A Short History of BPM, Part 8

Continued from Part 7.

Part 8 (the last): The Current State of BPM. Every analyst, vendor and customer defines BPM differently, because the current definition of BPM is very broad, and there are many vendors jostling for position within it. EAI/ESB-type vendors call their products BPM, but the products may contain only rudimentary human-facing functionality. Workflow-type vendors, also labelling themselves as BPM, lack the necessary infrastructure for integration, and often handle automated steps poorly. Some pure integration products call themselves workflow, just to confuse things further. There’s a lot of complementary products, such as process analytics and simulation, and business rules engines: BPM vendors will either tell you that a particular capability must be part of the base BPM product (if their product has it), or should never be part of the base BPM product (if their product doesn’t have it). And now there’s the whole SOA wild card thrown into the mix.

BPM is definitely a case where the whole is greater than the sum of the parts. It’s not just workflow plus EAI plus B2Bi plus business rules, plus plus plus: it’s the near-seamless integration of all of these tools into a single suite that provides an organization with the ability to do things that they could never do before. That doesn’t mean that all the tools have to be from the same vendor, but it’s essential to deliver all of the BPM functionality in a single environment of closed-loop process improvement.

Smith and Fingar’s book Business Process Management, The Third Wave describes this “third wave” as providing the ability to create a single definition of a business process from which different views of that process can be rendered and new information systems can be built. This allows different people with different skills — business manager, business analyst, regular old user, programmer — to view and manipulate the same process in a representation suitable for them and derived from the same source. They make a great analogy with HTML, where a business user may use a high-level tool like FrontPage to view and edit HTML, whereas a developer may edit the HTML code directly, but they’re still working from the same source. Round-tripping between a business analyst’s modelling tool and a developer runtime environment is one way to do this, although it violates the “same source” in the purest sense, but we definitely have to get rid of the strictly one-way paths from business analysis to implementation that exist now in many organizations.

Furthermore, Smith and Fingar point out that in the world of BPM, the ability to change is far more prized than the ability to create in the first place, and that BPM has the potential to actually remove application development from the cycle — the “zero code” Holy Grail that gets a lot of press these days. They make an analogy with VisiCalc, which took customized data analysis out of the hands of the IT department and put it in the hands of the business users, thereby taking software development off the critical path for achieving results.

Getting back to the point of this post, what is the current state of BPM?

First of all, we have several companies from the pure-play BPM/BPM suites market: they provide excellent human-facing BPM and at least adequate integration capabilities, with some providing outstanding integration. At the Gartner BPM summit earlier this year, they listed three “major players” in this category who had revenues upwards of $100M — FileNet, Pegasystems and Global 360 — and five “up and comers” with revenues above $30M — Appian, Lombardi, Savvion, Metastorm and Ultimus — while ignoring anything smaller than that. All eight of these vendors hit into the right zones in the Gartner and Forrester charts, which means that they either have the necessary functionality or are partnered with someone to provide it.

Second, we have a couple of integration-focussed BPM vendors who have purchased pure-play BPM vendors to create the complete range of functionality. The two highest-profile examples are the TIBCO acquisition of Staffware in 2004, and the BEA acquisition of Fuego earlier this year. In both cases, there seems to be a reasonable fit, but my concern is that the human-facing BPM side is going to become weaker since the main focus of these companies is on integration.

Third, we have the large software companies that have developed (or acquired) a BPM product: IBM, Microsoft and Fujitsu all spring to mind. In many cases, such as IBM and Microsoft, their BPM products are primiarly integration-focussed without a lot of human-facing support, and likely started as a “would you like fries with that” sort of offering for customers who were already committed to their architecture. IBM’s MQ Series messaging is probably still the most commonly used piece of integration middleware in financial services, although I think that they call it (and everything else) “WebSphere” these days, and IBM rightly has it as a cornerstone of their BPM strategy. Fujitsu is the odd one out here, with what appears to be a fully-functional BPMS; unfortunately, they’ve been marketing it in stealth mode and most people are completely unaware of it: as I said in one of my posts about the Gartner BPM summit, “who knew that Fujitsu did BPM?”

We’ll continue to see most of the business functionality envelope being pushed by the vendors in the first category as they seek better ways to integrate business rules, analytics, performance management and other capabilities into BPM; in fact, the most innovation seems to be coming from the smaller vendors in this category because of the lack of baggage that I discussed in part 7.

Because of the current focus on process improvement in all organizations, I don’t think that there’s any great risk of any of the vendors that I’ve listed here going out of the BPM business any time soon. However, the integration vendors will acquire some of the smaller BPM suite vendors to round out their portfolios, and the large software companies will acquire some of everything, in a continuing Darwinian cycle.

Before you vendors start adding self-promoting comments to this post, keep in mind that this is not intended to be a comprehensive list or review of BPMS vendors, and I know that you’re all very special in your own way. 🙂

The Eight — er, Four — Misperceptions of Outsourcing

I was catching up on some older Gartner podcasts recently — they’re not really time-sensitive, so fine to listen to them weeks or months later, and some of them do contain some good tidbits of information. There was one good one called The Eight Misperceptions of Outsourcing: Part I, in which Linda Cohen starts by listing these eight misperceptions:

  • the myth of sourcing independence;
  • the myth of service autonomy (this was particularly interesting since it touched on the subject of the interdependence of services due to SOA and BPM);
  • the myth of economies of scale;
  • the myth of service management as self-management;
  • the myth of the enemy;
  • the myth of procurement;
  • the myth of steady state; and
  • the myth of sourcing competency.

She then went on to discuss the first four in detail, whetting my appetite for Part II, which was to contain the second four. I checked my iPod: not there. I checked the iTunes directory: ditto. I checked the Gartner podcast page: Part II just doesn’t exist. Okay, it’s only been four months since Part I, maybe I’m being a bit impatient, but bring on the second four myths, already!

Of course, I’m not one to be throwing stones here: I posted the first six episodes of my Short History of BPM over a month ago, and haven’t completed the last two. Now that JC has caught up with translating them to French on his blog, however, I need to get moving on this.