Sunday, February 24, 2008
Got Money, Will Buy Sophistication
Thursday, September 13, 2007
I Can Give It Away Only If I Can Get It Back My Way !
At one of our clients, we moved to Salesforce.com – a SaaS vendor for sales information storage (things like price, purchase history, etc). A few questions arose immediately. Sure it’s less headache if the Salesforce.com manages sales related data. But what about the in-house ERP system needing it for order management purpose? How many other in-house or SaaS systems will need it later? What platform, formats, and protocols, should we support/address with perceived information exchange needs in future? And which of these systems is source of truth for a data element at a certain point in the lifecycle? While some such questions have generic answers, a significant chunk needs tailored solutions. SaaS vendors have way too many customers to create or support custom integrations for each one. The integration needs generated from “giving it all away” are too enormous for any SaaS vendor to take on. Oddly (or mystically) enough, just around the time that SaaS vendors started to take shape an answer to this problem also was brewing. Yup. If you are thinking, “this is an SOA Musings blog so the answer must be SOA”, you win our critical reading comprehension test!
Service Oriented Architectures and the standards that come with that provide self-service integration for customers of SaaS vendors. As long as the vendor provides standards-based interfaces, I can take those interfaces and publish data to the vendor or retrieve data out and massage it to any format I need. We recently did just this for the aforementioned client. This doesn’t mean there are no catches and gotchyas. There are plenty along the integration brick road but at least the heavy lifting is done and for the most part you are left with some XML manipulation, configurations, and business process development.
It’s no wonder SaaS vendors seem to be the stalwart of the technology sector. Look at the checklist for an IT manager:
* Easy to Manage: management handled by SaaS vendor
* Easy to Upgrade: Upgrades handled by SaaS vendor
* Easy to Ingrate: Standards-based integration can be done using SOA technology (however, in
general, no custom programming code to manage and maintain)
* Easy to Secure: YES (def. compared to 5-10 years back)
* Relatively Inexpensive: Yes
I saw how relatively painless it was to integrate with salesforce.com. Looking back over my 15 years in the industry, I can def. say that SOA has brought 3rd party integrations to a new level of ease and comfort. SaaS vendors would never have been the strong horses they are now before SOA. It was a really big deal to move company information outside the stone and iron gates of corporate
Saturday, April 21, 2007
Once Bitten, Twice Shy
At a client of ours, Web services were all the rage a few years back – maybe 4-6 years ago. Hearing about all the noise of the new paradigm, the IT dept there decided to embark on a Web services project as well. Busy minds worked head down researching the capabilities and the use cases that were associated with this and finally a proposal was put forth.
The proposal was essentially to offer various IT functionalities as Web services and then for IT applications to dynamically discover and bind to the most appropriate fit. Sounds vague and shocking? It was just as vague and shocking back then and the business teams at the client were just as skeptical. The project took two steps forward, five steps back; until finally it was laid to rest and Web services were forever labeled nothing more than "Hype".
Fast forward to 2007 and now we come in and propose SOA as the A2A and B2B integration substrate. And the feeling of being "once bitten – twice shy" was quite overwhelming. The reaction from both IT and business was exactly the same. “SOA!?! You mean Web services? Oh no, we tried that. It didn't work for us. It’s not for our type of business." And there began my efforts to explain the difference between Web services, the technology, and SOA, the uber-integration architecture. At the crux of my arguments was the idea that SOA is about integration rather than dynamic discovery. It's about loose coupling of integrated applications. It's not about dynamically choosing your business partners at run time. It took quite a few discussions for me to give SOA the confidence as the paradigm of choice. Our client was definitely skeptical but as of today they have a purchase order signed, sealed, and delivered for a SOA platform from one of the big guys and I sit with my soda and pizza and several architecture documents detailing the new SOA to be implemented over the next few weeks.
Tuesday, March 27, 2007
How many hammer strokes does it take to shape a rock?
We are all aware that any IT project is full of challenges. The ones that involve changing strategy are even tougher. But those related to SOA adoption are perhaps in a class of their own.
Enterprise architects in a company find it difficult to get the IT community rallied around SOA Strategy and Adoption programs. An EA manager recently lamented that as far as SOA is concerned, he would need to start looking at the calendar and "throw that watch away!" Sadly, he is not alone. Many strategist-practitioners have expressed dismay over how long it takes to move even an inch with SOA projects. One always wonders “Can’t we do it faster? Why can’t we just bring all people and all apps to SOA Wonderland in one go?”
The problem lies not quite with people resisting change, but more with people’s understanding of SOA. Some people simply want to put a SOAP façade in front of every application. These people consider SOA adoption as tactical issue not fitting in the strategy domain -– “Just create a timeline to add SOAP interfaces to all applications in the company”. Others ridicule the idea of finding a service on-the-fly and simply start using it -- “Companies would never do business like that!!!” Yes, those are things you hear when you ask about SOA, but that’s not exactly it. It’s not that straightforward or ridiculous. And this is not the complete list of misconceptions people have regarding SOA (another one: “I don’t want others to use my Web service until I can charge their location code).
In comes an SOA Strategist whose primary job is to do away with such misconceptions and apprehensions. Yes, SOAP interfaces, service discovery, reuse will all be involved in SOA adoption. However, a workable SOA adoption strategy involves creating (or adopting) a place for all of these, a context in which they are applicable and a process to do so. And this takes time. It involves qualifying service-worthy resources, creating service registration and governance processes, and a resource sharing and cost sharing agreements. Of course, not all such aspects are applicable in every SOA strategy. Nonetheless, these measures will ensure that the SOA adoption goals are not marginalized and people understand what it is and what it takes to make it happen. Taken together, each such step is building a robust SOA for the company.
Recently, while working with a client on SOA adoption, I discovered an excellent analogy for that. I’ve been working with an enterprise architecture group in a company that is putting forth a roadmap for IT to adopt SOA. Despite several slides, documents, POCs and use cases, there seemed little movement in an SOA direction. But just when everybody thought we should give up, a series of Aha’s led to the first ever SOA-based project to be implemented for the company. The IT director with the group shared a childhood story; one day he was watching his grandfather, a mason, shaping a rock by moving a big chunk off it. After that was removed, the grandfather asked, “So which stroke shook away the stone piece?”. “The last one of course” answered the young grandson. “Nope”, replied the grandfather “Every single stroke did it – a little every time.” What are your experiences with implementing SOA strategies? Is the mason right?
Wednesday, February 21, 2007
Are boutique Sis better at introducing new technology platforms?
I think it comes down to a few simple ingredients that the big companies can't replicate for each and every new technology because the risks are too high for them. What boutique SI offer is the following:
- Intense expertise in the new field
- either having invented part of it, worked on it from inception, or simply having earned the stripes through a few major projects
- Flexibility to work & invest outside a set project plan and strategies
- Lessened (yet loftier) expectations
- Boutique SIs often afford the luxury of having lessened expectations from a project management perspective. Large companies are expected to look, act, and behave a certain way; often tying their hands. Boutiques can shuffle around people from client to client and while there might be downtime here and there, clients tend to forgive and adjust more than with a Fortune 500.
- Don't let the lessened expectations lull you into a sopophoric state. Boutique SIs are not strong on the cash flow statement, they involve more risk, tend to be over all younger. There is an expectation from the client that the risk is made up for with a unique, cost-effective, and elegant architectural use of the new technology. The end goals are definitely loftier on the boutique SI.
- Boutique SIs are as famous as their last success; whereas larger consulting companies, independent or arms of software companies are assured some level of business because of their name and history. Customers benefit from the hunger to succeed that Boutique SIs are motivated by because they will find ways to make things work with available resources (people, h/w, technology, time, otherwise).
Tuesday, January 23, 2007
Network is the platform is the os is the app is the database is the...
The lights begin to flash wildly and the music starts to blare ending the crescendo with big bold Oracle logo right at the center and everywhere else. John appers, declares his allegiance to Oracle
partnership and jumps right into "making you uncomfortable about the future." He is making his point well about convergence of devices, hyper-personalization (my coinage for individuality of connectivity and content to max); but I'm still scratching my head -- Other execs from likes of Motorola are happy to give a sound bite in a 30 second video touting Oracle technology, but what makes John drive up all the way to SF from San Jose?
Halfway through the presentation, a pattern begins to emerge. Even though traditionally, Cisco has been a layer 2 to 4 (of OSI model) vendor, now things start to look more attractive on layer 7
(Application) level. During the demo, Cisco phones are sporting ball park e-tickets and discount coupons, Cisco video conferencing is helping solve crime on new TV series Vanished. The ubiquitous IP network (Cisco's traditonal playground) is of course making that happen, but it is taking the back seat in John's messaging. Hmmm... it's no longer just about the network.
On a slide animation, technical buzzwords rapidly begin fall off from application and OS layers down to the IP layer. Network is the next platform -- John declares and Cisco will take the leadership in making it happen (of
course). Essentially, what John was proclaiming that now Cisco also wants to be a serious player in application layer space. That should heat up the competition and thus it would be good for technology consumers. So what would it take for Cisco to make it happen. Throughout the glitzy presentation, John is alluding to benefits, we don't yet see the moving parts under the hood. And then I see it. On the slide showing top 5 priorities for Cisco, SOA makes its appearance! Now there is a big turning point for a technology.
We started with the application vendors like Microsoft, IBM, and Ariba evangelizing the technology, middleware vendors like Oracle and Tibco picking it up, through in-house development or acquisition or both, and now networking vendors are making it as part of their mainstream strategy.
Future should be good for SOA professionals. Viva SOA!
Friday, December 15, 2006
Ugly Ducking
Being the SOA strategist for the company and de facto architect of the very first SOA implementation, the ball fell in my court to find someone that fit the bill. "That should be easy." I mused. I just need to bring in some of the folks we work with if we are short on bandwidth in our company. Few phone calls and we have someone available! That in current state of the economy is an achievement! We talked about the whole arrangement and presented the person to the client. They all liked each other and everything was right on track -- until procurement got involved that is.
While talking to guys in the procurement maze, it became clear that they had a different view and take on the whole thing. The department had created a set of job profiles and related rate ranges. That sort of thing works well for most mainstream skill sets. It makes sure that the cost of certain skill set is streamlined across different departments. IT managers from all over the company can query procurement about what a typical, say, system administrator would cost. In this case, after the initial discussion, procurement folks concluded that this is a "software developer" requirement and thus falls within a certain rate range.
Now, I'm not going to judge how much they should pay to whom, but I can certainly say that SOA solution developers are not as plentiful as, say, Java or J2EE developers. While most of them would have done Java development in the past (and may still do as part of the SOA solution), it's not the same. Essentially, Java != SOA and C# != SOA and XML != SOA
To me, SOA solution development requires a different mindset. The way services are architected (message formats, granularity, composability etc.) is different from a Java application. Simply exchanging SOAP messages doesn't make a service. One needs to learn and internalize difference between SOA and glorified client-server application. And we all know, that more learned you are, the more price you can command -- And rightfully so. The more specialized a skill becomes, the more the demand per person there is and the laws of supply and demand say, higher the price.
Of course, not everybody has to agree with that argument. What's your take? Is SOA consultant the ugly (read pricey) duckling of the pack? feel free to add a comment or send email to naresh@tasmanAve.com.
Incidentally, our story has a happy ending. We have the consultant working on the project now.