SOA and the Data Centre
I was at an update on data centre strategies last week, covering such topics as virtualization and the concept of a shared IT infrastructure.
The thought occurred to me that actually there are a lot of synergies between data centre and SOA strategies.
It may seem bizarre to be linking what is essentially a software architecture from the world of applications and user programs with an area more concerned with hardware efficiency and utilization, power consumption and resource provisioning, but in fact these two areas have a lot of similarities. In essence, the shared IT infrastructure concept is about using data centre resources more efficiently. SOA is about using application and program resources more efficiently.
In fact, I believe that these data centre initiatives may be useful in addressing a potential future problem as SOA rollouts gather speed and become more extensive – the provisioning of SOA services. For anyone interested, there is a Lustratus Insight discussing the topic of SOA and the data centre in more detail, although I am afraid this is not one of our free papers.
It will be interesting to see how this develops, but it seems to me that companies that are really serious about SOA should start thinking about this subject now, in order to gain future advantage.
Steve
SOA market surveys: Buyers beware
Eric Knorr commented in InfoWorld a couple of days ago on Forrester’s latest market adoption estimates for SOA which showed much lower rates of adoption than the same survey suggested a year ago.
In my opinion, the key problem is probably not SOA adoption itself but problems with surveying end-users about hot technology areas. This is an area that I have recently covered in an insight called Interpreting Market sizings: It is a serious issue for organisations attempting to figure out what the real market take-up is and when or even if they should follow a particular technology trend.
Coming back to the article, Eric states that:
“SOA sounds great, but boy, is it hard. Especially on a wide scale, because doing it right generally requires rethinking how IT is organized.”
While I have some sympathy with the author’s sentiment, but Eric’s view really relates more to when and whether you decide to go for the wide scale big bang approach as opposed to the more achievable incremental build up.
More importantly, the evidence for this statement is the discrepancy between the 14% of respondents surveyed last year who said that they would implement SOA in 2006 (not including the 39% who already had) and the 2% who actually had a year later. This is a great example of the “Everybody’s doing it apart from me” effect. Asking aspirational questions about what we hope to do next year will always give misleading and optimistic answers. It is as true for healthy eating as it is for SOA! This effect is particularly strong with anything that is generally considered a hot topic (like SOA). This is of course not to say that the respondents didn’t genuinely think they were going to do a SOA project. But must-do projects appeared and other projects extended and then the wished-for SOA project was never got to.
What does all of this mean? It probably does mean that SOA take-up is slower than the more wild claims – no surprise there. It probably also means that people aren’t doing SOA as fast as they would like to – things always take longer anyway. However, it almost certainly doesn’t mean that there is a radical slowdown in actual adoption.
Finally… It is good fun to knock this survey and others (such as the one last year which claimed that 90% of companies would have exited 2006 with “SOA planning, design and programming experience” based on a sample of a massive 120 respondents). However, as I said at the start, the whole area of market surveys is a serious topic as it distorts perceptions both at the start and later on when more general adoption does happen. All of which means, with surveys – buyer beware!
Ronan
Is consolidation impacting on SOA innovation? Banking consolidation that is!
There has been a steady stream of SOA related start-ups being acquired over the last two years – most recently LogicBlaze by IONA.
Some commentators are beginning to suggest that SOA is fundamentally unsuitable territory for start-ups. I agree to a degree that SOA start-ups have a harder job today than their enterprise infrastructure software ancestors of 10 years ago – but I think it is less about SOA and more about changes in a key industry that used to support such start-ups: financial services.
To start with the argument raised by some against SOA start-ups such as Randy Heffner of Forrester who was recently quoted in SearchWebServices as saying:
a proliferation of small companies doing innovation it works against the goal of having a coherent platform with broad and rich functionality.
And
Innovation is one major value point that has to be balanced against the need and requirements of building a coherent software infrastructure platform for the next generation of applications,
I am far from convinced of this argument on two counts: if SOA is going to be truly a enterprise wide architecture it must span technology stacks and hence a single coherent software infrastructure platform may be hard to achieve and secondly SOA by its nature should ease the plug-and-play of products from different vendor into it. Of course, as Tony Baer of onStrategies points out in the same article, customers like ‘one throat to choke’ but they also like the solution least likely to result in a throat choking requirement.
Which begs the question why aren’t SOA start-ups succeeding at the rate one might have expected given the popularity of SOA. I suspect one contributory factor may be the continuing consolidation among the global investment banks based in Wall Street and the City of London. This was the traditional birthing ground for so many software infrastructure products and vendors. This consolidation means that there is much less room for SOA start-ups to build before breaking out into other industries. From my own personal experience of being involved with companies selling into the investment banks from the mid-nineties on, I can say that the task has got harder but more importantly the list of possible targets has got much much shorter.
Some examples: eleven years ago JP Morgan Chase Bank One was 6 banks (JP Morgan, Chase Manhatten, Chemical Bank, Banc One of Ohio and First Chicago NBD) of which at least 3 would have been on the target list of any start-up; Deutsche Bank took over Banker’s Trust in the late nineties; Union Bank of Switzerland merged with Swiss Bank Corporation to form UBS in 1998, in 2006 Mellon and Bank of New York agreed to merge and today JP Morgan is rumoured to be considering an offer for Barclays which is in turn attempting to merge with ABN AMRO – taking at least one more target off the list.
Of course, it can be argued that telecoms (another popular starting point for infrastructure software) has recovered in the last couple of years and other verticals could also pick up the slack with government in particular now a bigger buyer of integration products That is certainly true but government is often even harder and slower for a start-up to sell into than banking. Having managed to sell into both when I was running PolarLake, I am a good position to compare the scars!
Ronan
Software AG to become webMethods?
So, Software AG is buying webMethods.
I just hope this isn’t going to mean a hangover for users. The big question is: given webMethods is a more recognizable brand in the super-heated SOA market, will Software AG have the guts to take the bold step of renaming itself webMethods?
webMethods has long been a target of acquisition rumours. As the number 3 in the integration space back in the EAI days, it gradually lost pace with TIBCO and IBM, although it has a strong array of products. However, as more players came into what is now the SOA market, with strong moves from companies like Oracle and BEA, webMethods had been slipping further and further back. But the combination with Software AG is interesting to say the least. There is a load of overlap between the products of the two companies, so this will have to be resolved. Software AG made a move once before to get a major US presence with the SAGA initiative, only to back off when it was less than successful. Perhaps this is the major driver behind this move.
The future for end users will depend on the way Software AG jumps. In general, at least some of the webMethods products are superior to their SoftwareAG counterparts, particularly in the non-mainframe integration, process and BAM areas. Then there is the question of the salesforces. webMethods just rolled over a large portion of their sales force over the last 6-12 months, so this could be an area of even more churn.
However, what this does do is establish for arguably the first time a truly major European SOA vendor. Expect a more detailed advisory from Lustratus in the next couple of days or so, but for the moment our advice to users of products from either company is – don’t just look for confirmation that your product will continue, look for assurances that the competing product (and there is almost certain to be one given the extent of overlap) will be killed!
Steve
SOA: Slowly breaking the IT innovation bottleneck?
So often discussions about SOA return to the subject of how to communicate the SOA message to business, convince them to invest and to have faith in the opportunities that SOA can generate.
In these discussions, the CIO is often held up as both the potential champion and the potential bogeyman. A bogeyman in the sense that they are sometimes considered to be the blockers to innovation. SOA proponents sometimes mutter what some Web2.0 proponents shout from the rooftops. As Chris Anderson, inventor of the Web2.0 “Long Tail” concept put it recently:
“CIOs, it turns out, are mostly business people who have been given the thankless job of keeping the lights on, IT wise. And the best way to ensure that they stay on is to change as little as possible.”
Chris is right in pointing out that there is an underlying problem with enterprise IT and innovation. However, I think it is simplistic to point the finger at CIOs and their perceived defence of the status quo. Rather I believe Christopher Koch is much closer the mark in his excellent piece titled “Oh right, we forgot that CIOs are unimaginative as well as being fat, lazy and reflexive” (I couldn’t resist repeating the title).
“The predominant view, even in companies that claim IT to be “strategic,” is that the business owns innovation, not IT”
To put it another way: We may be able to reel off the IT successes that transformed the business from Walmart’s supply chain management to Fedex packaging tracking. However, these are exceptions rather than the rule. In many if not most cases, business managers simply do not believe that IT can be an engine of significant change and do not expect or even want IT to innovate. Partly, this is a cultural legacy with organisations not realising the degree to which their business is IT driven. It is also partly a cultural legacy of IT departments, only too willing to stay in their comfort zone. As Christopher puts it:
It’s time for CEOs and business management to acknowledge that just as their businesses are now completely reliant upon IT, they are more reliant on the geeks to help them innovate. And the geeks need to worry less about development languages and start worrying more about the business.
Obviously, it is also due to the technology legacy we are all familiar with which makes change hard and running to stand still enough of a challenge for many. To quote from Christopher again:
“As long as we persist in the naïve assumption that CIOs should be able to use 10-20 percent of their budgets to create a constant stream of breakthrough innovation on top of a creaky infrastructure that consumes nearly all of their management time and staff’s attention, we will not have significant progress.”
Is it all too bleak? Business regards IT as plumbers and IT is spending all of its time and resource patching leaky pipes. Of course, this situation is precisely that which leads to the enthusiastic acceptance of the SOA concept by enterprise IT: IT embracing the need to align with business and SOA providing the groundwork for a more agile environment and hence potentially facilitating a steady stream of small and well-targeted innovation. To quote Christopher, a final time:
“These small innovations add up, but more importantly, they can change more quickly, giving CIOs something they’ve never known with traditional applications and infrastructures: a measure of agility.”
This is precisely what we begin to see in organisations successfully moving along the SOA adoption path: CIOs, who have gained sufficient agility, can begin to play a role in driving innovation and a role in business strategy. However, this may take a long time in many organisations as it requires both delivery on the promise of SOA and a revolution in the perception of IT by business managers. And we must accept that during the process, there is no point pointing figures at the champion or other people who “don’t get it”.
Ronan
Mainframe SOA gets better
IBM has announced its latest version of CICS, V3.2, and the focus is definitely on making mainframe SOA better.
CICS is the mainframe transaction processing environment used by most of the Global Fortune 2000 companies, and it is estimated that trillions of dollars of investment are tied up in CICS applications. IBM started to deliver on its SOA promises with CICS Transaction Server V3.1, and now with V3.2 it has continued the process.
One clear attraction of IBM’s approach, for example, is the ability to have the mainframe play the role of both consumer and provider of SOA services. In other words, not only can CICS applications be turned into web services for external use, but they can also call other web services. This was delivered in CICS V3.1, and has been popular with a lot of CICS customers. The point is that, while being able to leverage CICS investments from outside increases the return on those investments, being able to utilize non-CICS services from CICS creates even more value and RoI. This bi-directional capability is something that many SOA tool suppliers ignore, seeing the mainframe as little more than a dinosaur, but it is in fact an extremely valuable function.
In CICS V3.2, IBM has improved its support for tools in an SOA environment, such as management, development and diagnostic tools, but at the same time it has tackled some knotty problems that have been hanging around annoying users for some time. The first is that IBM has finally acknowledged that IP is a key communications protocol, and it needs to be treated as such. So, V3.2 includes much better IP support. Another is the issue of CICS/MQ applications – CICS applications using the IBM WebSphereMQ messaging software, commonly used in SOA implementations where the mainframe is involved. In the past, the CICS /MQ adapter that enables CICS applications to use MQ has been owned by the WebSphereMQ product line. This has meant that MQ usage has always remained ‘outside’ the CICS environment, and support has suffered as a result. Now, this adapter has been brought into CICS. This has resulted in an immediate benefit of CICS making it thread-safe, improving its performance. Also, it can now be more closely integrated with CICS measurement and diagnostic tools.
Although users may greet V3.2 with cries of ‘About time!’, there is no doubt that the release does make mainframe SOA support even better. But what of all those users who cannot quickly get to CICS V3, and are stuck on V2 or earlier? Are they forced to wait for much-needed SOA benefits? Fortunately, IBM’s partner community has not neglected this opportunity – there are a number of SOA tools available from third parties to enable users to get started with mainframe SOA while they plan their eventual CICS V3 migrations.
Steve