IBM wants its SOA to be about two things: scaling and using technology to make the company money.
What is SOA?
Drew Cary is MC’ing the event. Looking up at the mega-screen after several muy caro jokes, he said:
I don’t even know what we’re sellin’. They’ve explained it to me 10 times, but I still don’t know. Yeah…’Smart SOA.’ Not that stupid SOA our competitors are selling.
That about sums up the The Emperor’s New Clothes side of the SOA-camp better than any other explanation we’ll probably come to this week. I saw excellent on IBM for being able to laugh at itself like this: it made the follow-up a bit more believable.
Makin’ Money
That follow-up was delivered by Robert LeBlanc, Steve Mills, Tom Rosamilia, and, of course, Sandy Carter. There was also a customer demo from Harley-Davidson’s CIO, Jim Haney. (Thanks to Dana Gardner for catching the line-up.)
In brief, as I Twittered, SOA is an idea above all else. An idea that states, the old tried-and-true enterprise IT motto of the past decade: align IT with business. LeBlanc gave a 10,000 foot definition of what this means that smacked of SAP-ery: all you tech people need to know your business’ KPIs, and make sure they keep doing well. Technology is just used to make money, the rest are details, as any Jonah will tell you.
As we discussed in a panel last week at MuleCon, getting tech people to take on that mind-set is tough. We’re not yet trained to think that way, and as I like to joke, those who do get promoted to management. There were no easy answers on the panel; we’ll see if IBM has some. LeBlanc did have a good opening motivator: be indispensable, not just irreplaceable. Don’t get off-shored, bub.
As an enterprise, the alternative to getting your own people to think like MBA-programmers is paying the high fees of outside consultants to do it. Sounds like good revenue if you can get it.
Also, as if by coincidence, see Galen Gruman’s story on a recent Hackett report:
Ironically, they achieve those financial efficiencies [generat(ing) $1.1 billion more operating profit on an annual basis and $645 million higher net profit] as an effect of their IT business value management approach, not from a focus on cost containment. Most companies — those that don’t get the strategic IT value — focus on maximizing the efficiency of IT, viewing it simply as a cost to be contained, said Eric Dorr, a senior business adviser at Hackett.
(I have no idea how credible Hackett is, but it fits well into the narrative from Impact.)
Scaling
The other angle hit-up by all was IBM’s differentiation here. The claim is that IBM can scale and SOA. One assumes both technologically and service/business-wise.
Here, there’s a mattress stuffed with numbers:
- IBM SOA customer China mobile adds over 100,000 subscribers a month with 99.5% customer satisfaction, 55 terabytes of data, and “19 million transaction per second telated to just billing.”
- An unnamed euro telcom uses IBS SOA – it’s ESB, as I recall – for over 14,000 service calls a minute with 100,000 completed order a day using “Native SCA” (?), Java, and CICS.
- IBM has had over 6,000 engagements around the world on SOA – one wonder about the criteria since SOA, really, is sort of “everything we call ‘software,'” maybe minus Lotus and Rational.
Scale. Yes, indeed.
Scale is the first defense/justification for any enterprise software vendor, so good job on throwing all that out. The key for incumbents – those who like Muffins and others – is to to dismiss scale as a just a commodity.
As IBM implies in all of it’s IT alignment talk (see above), scale is about more than transactions per a minute, it’s about being an enterprise software company (or partner) that support can its customers.
Big Shakes
Mills noted that there’s over 6,000 people here, which certainly seems the case. Paint me stupidly skeptical, but I’m astonished that over 6,000 people would turn up for a conference on SOA, WebSphere, and, you know, enterprise middleware, but here they are:
Disclaimer: IBM is a client and paid T&E for this event. MuleSource is a client as well.
Technorati Tags: ibm, ibmimpact2008, impact2008
"Native SCA” (?)"
In WebSphere Process Server, you have various ways your module can interact with dependencies. It could be a Web Service (SOAP/HTTP) call, a JMS queue, an EJB….Or, in the special case of one process server calling another process server module, you can use the "SCA Binding".
It's basically as fast doing a method invocation on a java class because SCA bindings (mostly) occur within the same JVM.
The reason you see "Native SCA" buzzword in the slide is to allude to how fast the binding is when compared to SOAP or JMS where you need to serialize.
Well we got 60 to an SOA conf in Sydney a few weeks back – must be IBM's pulling power ….
I'm not sure how much of a differentiator the 'scaling' aspect is … Oracle/BEA, ClearCape (?) and Sonic could make the same claims (and probably do!)
Keep plugging on that "SOA is an idea" theme … it's not a product set.
Dan: makes sense, thanks for the clarification.
Ric: yeah…it seems like closed source vendors, just by nature of usually having been around longer in this space, are the ones who make claims to scalability first-most, if not "second-most."