Wednesday, January 28, 2009

BriefingsDirect Analysts Discuss Service Oriented Communications, Debate How Dead SOA Really Is


Check out the full transcript at

Edited transcript of BriefingsDirect Analyst Insights Edition podcast, Vol. 36, on communications as a service and the future of SOA in light of hard economic times. [[and further edited/annotated in a gratuitous, unnecessary way by James Kobielus on the evening of January 28, 2009 for the purpose of populating this blog with sweet new stuff without having to work very hard]]

Panelists: Dana Gardner (Interarbor Solutions), Tony Baer (Ovum), Jim Kobielus (Forrester Research), Joe McKendrick (independent), Dave Linthicum (Linthicum Group), JP Morgenthal (Burton Group), Anne Thomas Manes (Burton Group), Todd Landry (NEC Sphere)


Gardner: I'm your host and moderator, Dana Gardner, principal analyst at Interarbor Solutions. Our topic this week, the week of Jan. 12, 2009, starts and ends with service-oriented architecture (SOA) -- dead or alive?

We're going to begin with an example of what keeps SOA alive and vibrant, and that is the ability for the architectural approach to grow inclusive of service types and therefore grow more valuable over time.

We're going to examine service-oriented communications (SOC) a variation on the SOA theme, and a way of ushering a wider variety of services -- in this case communications and collaboration services from the network -- into business processes and consumer-facing solutions. We're joined by a thought leader on SOC, Todd Landry, the vice president of NEC Sphere.

In the second half of our show, we'll revisit the purported demise of large-scale SOA and find where the wellsprings of enterprise architectural innovation and productivity will eventually come from.

We’ll also delve into the psychology of IT. What are they thinking in the enterprise data centers these days? Somebody’s thoughts might resuscitate SOA or perhaps nail even more spikes into the SOA coffin.


Baer: I hate to use a cliché, but it’s like the last mile of enterprise workflow and enterprise processes. The whole goal of workflows was trying to codify what we do with processes and trying to implement our best practices consistently. Yet, when it comes to verbal communications, we’re still basically using technology that began with the dawn of the human voice eons ago.

Gardner: I've seen people use sign language.

Baer: Well, that maybe too, and smoke signals.

Gardner: A certain finger comes up from time to time in some IT departments.

Kobielus: At least the use of a trusty index DTMS finger.

Gardner: There you go.


Gardner: Jim Kobielus, isn’t there more to this on the consumer side as well? We've got these hand-held devices that people are using more and more with full broadband connectivity for more types of activities, straddling their personal and business lives and activities. We know Microsoft has been talking about voice recognition as a new interface goal for, what, 10 years now. What’s the deal when it comes to user habits, interfaces, and having some input into these back-end processes?

An Important Extension

Kobielus: That’s a huge question. Let me just unpeel the onion here. I see SOC as very much an important extension of SOA or an application of SOA, where the service that you're trying to maximize, share, and use is the intelligence that’s in people’s heads -- the people in the organization, in your team. You have various ways in which you can get access to that knowledge and intelligence, one of which is by tapping into a common social networking environment.

In a consumer sphere, the thing is the intelligence you want to gain access to is intelligence sets residing in mobile assets -- human beings on the run. Human beings have various devices and applications through which they can get access to all manner of content and through which they can get access to each other.

So, in a consumer world, a lot of the SOC value proposition is in how it supports social networking. The Facebook environments provide an ever more service-oriented environment within which people can mash up not only their presence and profiles, but all of the content the human beings generate on the fly. Possibly, they can tag on the fly as well, and that might be relevant to other people.

There is a strong potential for SOC and that consumer Facebook-style paradigm of sharing everybody’s user-generated content that’s developed on the fly.


Text-Mining Capability

Kobielus: One of the services in the infrastructure of the SOC that will be critically needed in a consumer or a business environment is a text-mining capability within the cloud. That can go on the fly to all these unstructured texts that have been generated, and identify entities in relationships and sentiments to make that information quickly available. Or, it can make those relationships quickly available through search or through other means to people who are too busy to do a formal search or who are too busy to even do any manual tagging. We simply want the basic meanings to just bubble up out of the cloud.


Kobielus: I want to add one last observation before we go to the "SOA is dead" topic. In order for this integration to happen in the cloud, the cloud providers need to federate their new registries with those of their enterprise customers. But, humans are reachable through a different type of registry called a directory, lightweight directory access protocol (LDAP) and other means.

Cloud providers need to federate their identity management in a directory environment with those of their customers. I don’t think the industry has really thought through all the federation issues to really make this service oriented, business communications in the cloud scenario a reality any time soon.

Gardner: So we need an open Wiki-like phone book in the sky.

Kobielus: Exactly.


Kobielus: The whole "SOA is dead" theme struck a responsive chord in the industry, because there's a lot of fatigue, not only with the buzzword, but the very concept. It’s been too grandiose and nebulous. It’s been oversold, expectations have been built up too high, and governance is a bear.

We all know the real-world implementation problems with SOA, the way it’s been developed and presented and discussed in the industry. The core of it is services. As Anne indicated, services are the unit of governance that SOA begot.

We all now focus on services. Now, we’re moving into the world of cloud computing and, you know what, a nebulous environment has gotten even more nebulous. The issues with governance, services, and the cloud -- everything is a service in the cloud. So, how do you govern everything? How do you federate public and private clouds? How do you control mashups and so forth? How do you deal with the issues like virtual machine sprawl?

The range of issues now being thrown into the big SOA hopper under the cloud paradigm is just growing, and the fatigue is going to grow, and the disillusionment is going to grow with the very concept of SOA. I just want to point that out as a background condition that I’m sensing everywhere.


Kobielus’ comments on all the above, here on the evening of January 28 by himself:

• “Trusty index DTMS finger”? What does that mean? I never said it. Whoever transcribed the audio misattributed it. What does “DTMS” stand for anyway? Go listen to the audio playback and tell me whether I actually said it. I’m too lazy to do so. Also, I can’t stand listening to my own voice (yeah, I know, you’d think otherwise, wouldn’t you, given how verbal I am).
• Mash up each other’s presence, content, intelligence? A virtual mashpit, so to speak. Feels slightly creepy, doesn’t it. Sort of like the movie “The Fly” where the machine went haywire and mashed Jeff Goldblum’s DNA with an insect. Ewwww!
• Glad I mashed up identity management and service governance on the call--mashed and mushed LDAP directories and UDDI registries, into each other conceptually--and federated them in that big Venn diagram in the sky. SOA for interpersonal communications depends on populating the governance bus with all that identity “metadata” (e.g., contacts, attributes, profiles, roles, demographics, interests, transactions, behavioral characteristics, clickstream, predictive model scores, etc.)
• Text mining will provide the auto-discovery mechanism for all the “identity metadata” that people are self-publishing in un- and semi-structured formats (often without fully realizing it) in the Web 2.0, social networking, wiki world.
• Controlling mashups. Mashup governance. Dave Linthicum introduced that concept a year or two ago, but I still don’t sense any clear feeling among vendors or users that it’s a hot button. I think everybody still regards user-created services (i.e., mashups) as outside the proper scope of SOA. But, then again, what’s the difference between a mashup and a rogue service? The former is not sanctioned by corporate IT but is ostensibly benign and is to be tolerated, if not encouraged or supported. The latter is also unsanctioned, and possibly benign, but under suspicion and to be decommissioned or neutralized at the first opportunity. And what’s the difference between mashups and virtual machine sprawl? The former proliferates but doesn’t necessarily hog resources or disrupt operations, whereas the latter also proliferates and consumes more than its fair share of resources. The relevant distinctions in these cases all concern where specifically a particular created/published service (be it a mashup service, Web service, or cloud service) sits on the governance spectrum in a given organization. Is it a sanctioned/supported or unsanctioned/unsupported service, from the point of view of the service governance “authorities”?