It was that experience that convinced an enterprise architect at Caterpillar to implement portfolio management tooling – a decision he credits with greatly improving IT’s ability to respond to initiatives and a significant labor saving success.

Mark Zielinski, strategist and enterprise business process architect at Caterpillar, explains that the seriousness of the challenge hit home when he was asked to provide a business partner with a comprehensive inventory of applications in use at a model manufacturing facility, and couldn’t. According to Zielinski it took an architect working on site for 30 days to ferret out all the bits and pieces and deliver a complete report. Zielinski spoke as part of a recent Troux Technologies webcast that covered portfolio-related issues.

Zielinski and crew wanted to be sure they could inventory the applications. As a result, Zielinski noted, he and Caterpillar implemented Troux Enterprise Portfolio Management. That software now provides visibility into more than 2000 applications and allows Zielinski immediate access to the mix at any given Caterpillar site.  He compares the “heat map” produced by Troux to a view of the company’s DNA.

He credits that capability with saving labor and speeding the process of developing reference architectures.  Longer term, Zielinski says the information gives product owners and architects the detailed information they need to support transformation initiatives.  It also has a role in application rationalization and as a “facility configurator” when new sites are brought online.

Thus, Caterpillar – a giant, growing multinational – arrived at a situation where its ability to engage in meaningful transformations – of the business or of IT – are better enabled.

- Alan Earls, Contributor

Today, ”the parts are interoperable,” he said, noting that ”APIs have now become products for a lot of companies and public sector agencies.” These public APIs must be flexible, and be able to support an increasingly broader user base.

”APIs are now recognized as a first-class revenue generator that is solidifying the need to have a service-oriented enterprise – one with the expertise to do services and to be agile,” he said.

Gordon said Unisys is rolling out new services, known as the AMPS Center of Excellence, to help companies improve their application modernization initiatives. The services suite includes an AMPS SOA Governance and an AMPS SOA Operational Software Platform. Some of the new parts are supplied via deals with other software providers such as EMC Documentum, SOA Software and RedHat JBoss – with special Unisys-tailored customizations, based on the company’s extensive work in the field.

SOA infrastructure provides a useful ”backplane,” according to Gordon, to help orchestrate and manage the new style of API. ”A SOA management intermediary is valuable,” he said. ”It brings the management of APIs, security, logging, protocol mediation and a dashboard for watching services activity.”

Still, like others, Gordon emphasizes that ”SOA is something you do, not something you buy.” As a result, SOA Assessment services and SOA Strategy services are part of the AMPS Center of Excellence.

We asked Gordon to share a few useful tips for achieving a successful enterprise SOA. He noted three elements that need to be in place. These follow.

3 Tips for Laying the Strategic Groundwork of a Successful Enterprise-Wide SOA

1. A prioritization process for requirements that emphasizes enterprise priorities in lieu of departmental priorities

2. A highly transparent, participative governance process comprising all stakeholders including a virtual team of service providers led by the SOA Program Director to ensure uninterrupted support for the SOA initiative.

3. An unbounded commitment of an executive sponsor to steadfastly support this organizational transformation.

Setting Priorities
Requirements for services in an enterprise-wide SOA initiative are determined and funded according to the priorities of the enterprise as a whole, rather than those of departments. This forces alignment of business with IT ensuring the goals of SOA are aligned with organizational objectives. That requires participatory governance and communication processes and especially greater interaction with the business lines.

Governance
Sound governance begins with a strategic plan that includes the business goals. These goals, in turn, can be transformed into IT requirements with clear line of sight from business goal to IT requirements, followed by high-level design specifications through testing, deployment maintenance, and application end of life.

Executive Commitment
A key responsibility of the executive sponsor (or their delegate) is to be the final decision making authority when the participative governance process reaches a stalemate during a task. The stalemate may occur during requirements prioritization, or there could be a disagreement on the timing for delivering new capability to customers.

When we spoke with MIT Systems Researcher Jeanne Ross, she pointed to Boston’s Big Dig as an archetypal muffed project.

This was a massive federal project to replace Boston’s Central Artery, which added a tunnel to the airport and replaced an elevated highway with a depressed one. It started in earnest in the late 1980s or early 1990s, depending on how you estimate “earnest.” Originally set for 1998 completion and budgeted at $2.8 billion, the eventual cost was over $15 billion (by some estimates, the final tag was $22 billion) and its completion occurred late in 2007.

From Ross’s point of view, The Big Dig was an example of classic underestimation of the kind that can happen in software development. How much are we spending on an existing system while we are laboriously rebuilding it? Some people don’t deal with those figures. Here is Ross’s take:

I highlight the Big Dig here in Boston because the thing cost over $15 billion dollars. The idea was: “We’re going to take this highway that goes over the city, and we’re going to put it under the city.” Brilliant idea! The main thing they underestimated was what happens to [existing systems] while you are building the new capability. They were just trying all kinds of new technologies and approaches to doing things, and they really had no way of estimating how long it would take or how much it would cost.

The lesson in it for architects in IT is that at least a third of the $15 billion was spent on just making it possible to continue to do business, for cars to continue to be able to navigate their way to the airport, to the office, wherever it is they’re trying to get to, even while this major transformation was taking place downtown. So, a third of the money is not going towards the new capability, it’s going towards being able to continue to exist while you’re building the capability. I think that’s a thing as a rule that architects would like to shy away from. It feels like such a bad use of your money.

To read the rest of Ross on business capabilities, go to “MIT’s Ross says exploiting IT business capabilities is next EA step.”  Ross, director and principal research scientist at the Center for Information Systems Research (CISR) at the MIT Sloan School of Management, spoke with us shortly after speaking at The Open Group Conference in San Francisco. – Jack Vaughan

Today, Accenture released an updated version of a self-service portal to provide people electronic access to their caseworkers and the benefit application process in state resource systems.

Accenture Citizen Self-Service Portal Version 2.0 has enhanced self-service features, a training module and a ”natural language” rules engine to increase public agencies’ flexibility to adapt benefit eligibility programs and processes in response to legislative and policy changes.

Human service agencies using the portal will be provided ”24/7” access to citizens who want to determine their potential eligibility for public assistance programs, apply for benefits and manage benefits.

The new portal is designed mainly to help meet the needs of social service agencies and the people served by those agencies. It builds on the Accenture Public Service Platform to help public agencies manage human services delivery cost effectively with improved technological flexibility and enhanced citizen services and program outcomes.

This is part of an on-going Accenture effort to provide SOA expertise to government agencies. Read here to find how the company is now working with the state of Kansas to help create a new benefits eligibility system that implements service-oriented architecture.

It is hard to ultimately succeed with SOA services unless you take the wider view. That view must include an understanding of the organization’s overarching goals.  SOA has a technical angle, yes. But too often, SOA advocates have sent the business owner into trances with technical particulars – WSDL handshakes, ESB performance tuning metrics or service normalization patterns.

A shared vocabulary is what is needed, writes William Ulrich, head of TSG consultancy, featured in a recent SearchSOA article on application modernization issues. He sees the term and practice of business architecture gaining definition. Check out ”William Ulrich on ‘Business Architecture’- Seeking a common language.” – Jack Vaughan

Larry Turley, the president of Ron Turley & Associates Inc (RTA), explains that right now his people are nearly 100% COBOL, but eventually he thinks it will switch over to mainly Java. Maybe not during his time as president at the company, but it will likely happen.

In the meantime, Turley and his COBOL developers are finding success implementing Java functionality with a COBOL modernization platform—isCOBOL Evolve from Veryant—that lets them write their code in COBOL, but deploy their apps in Java. RTA had been using ACUCOBOL as their development platform for over 20 years, but when that company was acquired by Micro Focus, some of the changes made them feel it would be better to take a look at what other options were available.

According to Turley, there aren’t that many options for PC developers using COBOL. There’s still a lot of support for the mainframe and midrange folks, but when it comes to a strictly PC-based clientele—which is RTA’s bread and butter—there’s just not much support.

COBOL does not have a reputation for being as Web-friendly as Java or .NET languages. So it follows that there’s a lot more focus on these newer languages. And even a 30+ year veteran of the COBOL wars like Turley has to admit that being Web-enabled is a huge advantage in today’s application development environment. In fact, building Web-based applications is the single biggest factor driving Turley’s decision to start deploying in Java.

According to Turley, Web services are definitely going to be a piece of that. He says that, “The hope is that we will have a lot of service-based interactions. Some of those will have to be developed as we go.” He plans to build out Web services as clients need them for particular purposes and use those experiences to build a library of practical reusable Web services.

Forrester analyst Phil Murphy pointed out that one size does not fit all when it comes to mainframe legacy modernization. Some mainframe code is not so old that it does not have some flavor of object-oriented software. Some code is in good shape, and thus, given your overall strategy and compute horse power, it may be an economical candidate for re-hosting. The question to consider is whether you have a well-structured application, noted Murphy. Read ”Modernize? Consider the MIPS” on our site.

Why does the corporation look so favorably on cloud computing? It seems to smell like cost reduction. But ROI studies do not always confirm the cloud savings. Are IT departments doing thorough ROI studies on cloud? Fewer than you might expect, according to Chris Harding of The Open Group. ROI should be measured, and the variability of your processing load is a matter to consider as you do those measurements, said Harding. Read about ”Cloud computing myths and the developer’s role” on SearchSOA.com.

“If you have a fairly steady processing load, cloud doesn’t make much sense, but if you have a variable load it can be appealing,” he says. However, Harding theorizes that respondents from larger organizations may see enough variability within business units and departments, for example, that the overall load balances out – making an ideal environment for private cloud. “That will probably be cheaper than going to an external cloud supplier,” he adds.

We spoke as well with analyst Judith Hurwitz. The topic again was cloud computing. She urged users to consider that some of this is new, and some is not.  As teams move their middleware work to the cloud, they find the fundamental rules apply – that building out application servers remains part of the job. It is, in fact, sometimes a most tedious part. “Even though it is on a cloud, the issues of enterprise development are still there,” said Hurwitz, pointing to configuration and metadata handling as examples of such issues. Read ”Middleware in transit” on the site.

Cloud computing was also front and center at last week’s IBM Impact 2011 event in Las Vegas. IBM announced moves to streamline provisioning of middleware images on cloud nodes. This is an important step. The first cloud applications, with some notable exceptions, were pretty much straightforward number crunching apps — one-trick ponies, if you will. But, increasingly, the apps on the cloud will come to mirror the complexity of “apps on the ground,” ones that use sophisticated middleware. Read all about it in our Impact 2011 show story.

Clearly, the cloud was out in force last week at SearchSOA.com. Alan Earls contributed a piece that looks at the future of midrange legacy systems and the notion of cloud migration. Earls’ effort is posed as a question: “Is it time to migrate your midrange assets to the cloud?” He suggests that the vestigial midrangers are as hard to decipher as mainframes. That and other issues may slow any effort to move the midrange legacy to a new platform.

According to Uppal, there are four key points to realize in order to successfully integrate capability-based planning with your enterprise architecture.

First, combine TOGAF with capability-based planning to develop a closed loop value chain process that integrates existing disciplines with respect for each discipline and without creating a hierarchy of disciplines. Next, recognize that IT is a line of business (LOB) and as such is responsible for maximizing the use of current capabilities to achieve an acceptable service level at an acceptable cost.

Also, you must have a considerable focus on acquiring new capabilities, but make sure that new and old assets are integrated properly to maximize the capabilities of both.

Finally, remember that each LOB is responsible for delivering particular services and that they will define how said services are delivered – the role of IT, according to Uppal, is to aid other LOBs in defining their services effectively and efficiently, both for the individual LOB and for the greater organization.

A little while ago, we had the great pleasure of hosting a chapter excerpt from modernization mavens William M. Ulrich and Philip H. Newcomb. If you want a quick view on modernization pitfalls and strategies, their book on systems transformation is an excellent place to start. Meanwhile, some other of our recent coverage on this topic has been aggregated in our special report on mainframe application modernization.

Cloud computing on one level is a variation on grid computing, a technology this site began to cover about 10 years ago. APIs are where the pedal meets the metal in the cloud, and we looked at these in a Cloud API mini-roundup a while back. Meanwhile, the data requirements of today’s cloud projects appear to be quite novel, and they are the area of interest for quite a few services architects these days. Last week, SearchSOA.com’s James Denman uncovered some useful resources on the graph database, one of the extra tools in the NoSQL movement which is poised as an alternative to the traditional relational database in cloud and other apps.

It wasn’t too long ago that EAI and SOA pioneer David Linthicum discussed Web data services and distributed computing with us. Sometimes the rush to cloud belies the role services play. Linthicum’s discussion is an antidote to that. Also, after a bit of hiatus, expert Michael Ogrinz has checked in with a very interesting look at data mashups that employ government data services. It is another take on the emerging realm of data services, discussing tools and techniques and using a ready example.