 | Level: Introductory Grady Booch, IBM Fellow, Rational software
15 May 2005 The change in leadership of the Rational brand, from Mike Devlin to Danny Sabbah as general manager, gave Grady Booch an opportunity to discuss with both leaders their goals, accomplishments, and vision in the field of software development. These interviews provide insight into Rational's history as well as the future of IBM's software development brand.
After twenty-four years at the head of the organization that put iterative software development practices on the IT map, Mike Devlin, general manager of IBM Rational software, is retiring. Taking over the Rational GM position is Danny Sabbah, formerly the VP of development for the WebSphere brand, and more recently the head of Strategy, IBM Software Group.
Mike Devlin co-founded Rational Software Corporation with Paul Levy in 1981. The company set out to create a commercially available environment for improving the processes associated with software development -- a goal that led to the Unified Modeling Language (UML), the Rational Unified Process (RUP), and a host of tools that facilitate today's accepted best practices in software engineering. As a mark of the company's success, Rational was purchased by IBM in 2003 and is now one of the five major brands within IBM Software Group.
Danny Sabbah joined IBM in 1974, and earned a PhD in computer science in 1981. Before moving into IBM Software Group to become VP of architecture and tools development, he worked in telecommunications, artificial intelligence, programming languages, and software technology. In the late 1990s, he led the team that delivered the WebSphere Web application server, and as the VP of development for the WebSphere brand, he was responsible for the architecture and strategy for IBM's application development tools, as well as IBM's Web and business integration software.
The passing of the Rational baton from Mike Devlin to Danny Sabbah, both considered visionaries in the software development arena, suggested an historical opportunity to interview the outgoing, as well as the incoming, leader of IBM's software development brand. And who better to conduct the interview than Grady Booch, who has known Mike since their time together at the US Air Force Academy, and Danny since nearly the beginning of his distinguished career at IBM?
Grady's question-and-answer sessions that follow illuminate not only the personalities of both leaders, but also the path that Rational continues to blaze through the complexities of software development. I think you'll appreciate this look at the past as well as the future of Rational software.
--Mike Perrow
Questions for Mike Devlin, retiring as general manager of Rational software, IBM Software Group
Grady Booch: Rational has pushed modern iterative software processes for years. Do you think iterative development is mainstream today?
Mike Devlin: Today iterative development is generally accepted as the industry best practice. Very few customers would disagree with the goal of achieving iterative development, since it offers teams not only a sound approach to new system development, but also a process for incrementally improving existing software systems. However, only about one third of our customers actually practice true iterative development. A major part of our value proposition over the last twenty years has been to help customers move to transform their development process to adopt more modern methods, such as iterative and asset-based development.
GB: What impact did Rational have on the software process world? Did RUP shape the integration of tools or vice versa?
MD: Rational was a thought leader from the earliest days and had a great impact on the direction of software engineering. Back in the early and mid-1980's many of our customers were in the Aerospace and Defense sector. Introducing iterative development was a challenge given all the cultural change required. At that time iterative development actually stepped outside the Federal Acquisition Regulations and associated FAA and DoD standards that enshrined the Waterfall lifecycle model (and they enshrined functional decomposition, making it difficult to apply ideas like data abstraction, information hiding, encapsulation, and modern OOP). Working with key customers we eventually were able to change the standards to actually encourage iterative development and modern architecture.
Similarly, in the early days of Object Oriented Analysis and Design we were faced with a fragmented market with conflicting notations and standards. We orchestrated an effort to achieve a single industry standard, which culminated in a specification for the UML. The first step involved bringing the thought leaders (who were also the market leaders) into Rational in the form of the "Three Amigos": you, Grady (you were at Rational from the beginning), Jim Rumbaugh, and Ivar Jacobson. Even more challenging was gaining support of all the key industry players (Microsoft, IBM, HP, etc.). Once we consolidated around a single direction for the UML and achieved market leadership, it was possible to establish a formal OMG standard.
The RUP provides a framework for capturing all of our experience and best practices. Then by supporting RUP in the tools we could enable customers to implement those best practices. The RUP definitely influenced our tools, but it is also true that the pragmatics of available tool technology influenced the RUP.
GB: Given Rational's history of support for multiple platforms, what attracted you to IBM?
MD: Customer requirements drove Rational to support multiple platforms (including others besides Microsoft and IBM). However, we needed to move to a single technical architecture to reduce development costs and allow us to move more rapidly in the market. Only IBM provided an open technical architecture able to support multiple platforms and to allow us to accelerate our development activities. So the choice was clear -- moving to IBM was the right thing for our customers and for our business.
GB: Why was "customer success" the overarching philosophy in your selling model and how did you anchor the company's culture in that theme?
MD: When we started Rational it was very clear to us that customer success was the core of our business, almost a fundamental integrity issue. We felt that our technology could have a huge impact on the economics of software, but only if customers used the technology successfully.
It was also in our self interest to ensure customer success. Our early customers were large Aerospace & Defense contractors and large Telecom Equipment companies. These are both concentrated markets where any failure to perform would quickly be spread by word-of-mouth throughout the industry. As a small company it was a survival issue for us to ensure customer success, and therefore it naturally became the centerpiece of our culture and our way of doing business.
The early customers reinforced this by frequently pointing out that our people and our expertise were at least as critical to their success (and continued purchase decisions) as the tools themselves. This led to our model of having three technical people in the field (including what is now called lab services) for each sales person. Every manager understood that successful customers generated more business at a lower cost. Once again our customer's success was aligned with our own success.
To institutionalize this, we first built a (simple) field measurement system where the first (of five) metrics was customer success. This was a subjective assessment by managers, but since field managers were close to most accounts, it was easy for them to get direct feedback.
Then we made sure that Marketing and Engineering understood that their job was to make the field organization and the customer successful. All other functions (HR, Finance, etc.) were subordinate to those functions (Field, Marketing, Engineering) that served customers. Throughout the company we celebrated customers and the teams that served customers.
GB: An early Rational technical notion about the software development environment was "keep the intermediate representation persistent." Was this the beginning of Rational's belief in version control, or other concepts ancillary to the code lifecycle?
MD: Our early experience with a rich intermediate representation (IR) convinced us of the value of true semantic integration within the tools. Having the IR allowed us to achieve a much higher level of integration and to provide much greater automation. Today we achieve this with Eclipse EMF.
This experience also taught us about the important interactions between architecture, configuration management, tools, and process. Clearly, semantic integration introduces notions of semantic consistency, which have implications for configuration management and version control. However, there are even deeper issues around the application architecture and tools support. We discovered that our tools could encourage and support well-constructed (componentized) application architectures. However, we also discovered that poorly architected applications create major configuration management problems and do not leverage the full power of our tools.
GB: Given that your work spans the decades as well as the globe, how would you characterize how software has grown in importance and how the process of developing software has changed?
MD: Our vision in starting Rational was that all sectors of the economy, particularly the high growth sectors, would become increasingly dependent on software. Therefore software development was the most important technology arena in the new post-industrial, information economy.
The reality of the world today demonstrates the validity of our original vision. Most products today (telephones, cameras, stereos, automobiles, aircraft, medical devices, etc.) contain increasing amounts of software and are designed and manufactured using software systems. Most services (financial services, distribution and logistics, retailing, entertainment, etc.) are based on software.
Software (and the IT industry) is also credited (at least in the US) with driving high growth, high productivity, and increased workforce participation. Around the world, software has accelerated globalization, which has brought increased prosperity to developing, as well as developed, countries.
While the importance of software is now widely accepted, the development process has not evolved as rapidly as we might like. Improved development process, more automation, increased collaboration, global sourcing, common standards, common architectures, open source, and other developments have greatly improved the economics of software. This has enabled the proliferation of software throughout the economy.
However, we all know that most software development is still very challenging, with many projects that are late, over budget, and responsible for software that is not of high quality. As an industry we need to continue to raise the level of quality and productivity even higher. IBM is in the best position to do this, given that only IBM can provide all the technology (tools, processes, architectures, all based on standards) an open framework facilitating collaboration with customers, partners, industry organizations, academia, the open source community, etc.
GB: What was the hardest part of making Rational successful? Was it the technology itself, was it the problems of development, or was it the challenges of the field and delivery?
MD: It is hard to single out one piece of the whole as being "hardest." It seemed to me that everything was a critical success factor. The technology was very critical. Getting the first release out was more difficult, with higher risk and longer development time, than expected. Each release since then has had its own challenges as our customer base grew and our technology evolved. It was also very difficult to build a field organization that was deeply technical, was committed to customer success, and could consistently deliver business results.
The key to our success in these areas was the team and the culture. We carefully selected only those people who could bring high quality and commitment to the mission, and who were compatible with the team-oriented, customer-oriented culture we were building. Once we built the right team, everything else was possible.
GB: In your years with Rational, what was the most fun part of the journey?
MD: Without question the most fun AND the part I am most proud of is the team itself. We built a great organization, one that really understood the importance of team over individual play. This was a team that really believed in the mission of ensuring customer success, and which took personal responsibility for every aspect of the business. We produced results, primarily because we as a team established and maintained the right culture.
GB: What advice would you give to someone -- perhaps a high school or college student -- about a career in software?
MD: Advice? Nothing fancy. Work hard, set impossible goals, stick to your principles, and deliver results for your customers. If you can do these things consistently, you will be rewarded in many ways. You'll make a difference to your teammates and your organization, and you'll very likely make a difference in the IT industry. And yes, there's financial success as one of the rewards. But just as important is your personal and professional growth. There's nothing more satisfying than creating something of lasting value.
Questions for Danny Sabbah, incoming general manager of Rational software, IBM Software Group
Grady Booch: Mike Devlin and I have known you for years, as have many of our customers and colleagues in the field of software development. But for readers who might be less familiar with your work, could you describe your background in software development?
Danny Sabbah: Sure. When I was in the research division, in the mid to late 1980s, I headed a team called "Software Technology, " which focused on programming languages and development environments. By that time, I had already developed a deep and abiding interest in core software technology and programmer productivity, and in understanding the day-to-day life of the developer. When I began working for Steve Mills in IBM Software Group [SWG], I spent the first three years focused solely on IBM's application development business. To complement what we were doing in application development, we kick started the WebSphere effort, which was late 1996. Of course, in the early 90s, we had lost the application development space to VisualBasic, to Borland, a few others. But we had, and still have, a strong market presence regarding language environments, like COBOL and PL/1, and that's now matched by Java. So a major aspect of our Java initiative within WebSphere -- including WebSphere Studio, as it was formerly called, and the Eclipse initiative that also came out of my team -- was to rekindle our ability to influence and grab mindshare in the developer space.
GB: Why do you think software development tools are so important to IBM? Do they really require a separate, high-profile brand in the IBM software portfolio?
DS: I've always believed that software development is critical to any organization that wants to succeed in a modern business climate. One of the main reasons IBM decided to acquire Rational -- and I'm proud to note that I was on the decision team -- was to fill out and complement what we were building regarding an integrated development environment. We needed lifecycle capability. No developer manages an individual piece of software without understanding how to do everything -- from gathering and managing requirements over the lifetime of that software, including maintenance, to managing the modeling and coding of the software; from versioning code, to testing it throughout the lifecycle. All these things are crucial. An individual IDE is not terribly useful without the lifecycle management and modeling artifacts.
As a major brand, Rational is helping us establish and maintain mindshare and market leadership around the discipline of software development, around creating and managing software assets in the enterprise. This is why it made sense last year to move some of the development tools from WebSphere into the Rational brand. Basically, we made a conscious decision, John Swainson and I, that to better integrate Rational within the IBM software family we needed to move the development team as well as the development assets -- from both Eclipse and WebSphere Studio -- into the Rational brand.
So to a certain extent, you can view my transition to Rational GM as a reunion with my past as well as with some of the folks who worked for me in the past. The complement of the overall Rational toolset is something I've been looking forward to for many years.
GB: Can you give an example of how Rational's software development tools and best practices are helping IBM's software initiatives?
DS: Here's a great example. Rational is already having a positive impact regarding our commitment to Web services and service-oriented architectures. When I was in Strategy, I started an evangelism effort around SOA, and a large part of what we did was to work with the Rational team to create development tools that allow developers to understand asset lifecycle management -- not only inside IBM, but also with our immediate customers. More broadly, the intent is to capture mindshare and market leadership around regional system integrators, as well as the average developer in custom development shops. Services and SOAs will not be successful without a full complement of development tools, many of which need to come from Rational. Certainly the guidance and leadership in this market space needs to come from Rational.
My view is that Rational plays an important role in determining success in SOA, a role that's as key, if not more so, as the role WebSphere plays in supporting the runtime environment. Because if you don't capture the developers, then the runtimes are basically secondary.
GB: How is Rational contributing to the growth or acceptance of the enterprise service bus model?
DS: Software Group's notion of the enterprise service bus is that it's not an end in itself, but a means to an end -- a way to craft looser coupling between self-defining interfaces. These self-defining interfaces are defined and managed in an asset lifecycle management model that needs to come from a modeling discipline, and a set of tools that come out of Rational. The methodology elements are extensions to RUP, so many of the technologies that we've been adopting through our adherence to RUP come directly from the Rational toolset. So whether it's the deployment of ClearCase, the adoption of WebSphere Studio (now called Rational Application Developer), the incorporation of portlet builders from Rational toolkits -- all of these are being used internally and are helping us create and isolate our assets that then become core components in the future development of our products.
For our customers, just as Rational has promoted technology shifts in the past around RUP -- for instance, marrying some of their tool capabilities to an integrated view of how you do object-oriented design -- the same thing needs to happen around SOAs and service-oriented design. The creation and management of assets is just the next stage in an evolving modularity for our development tools, methodology, and runtime platform. So the notion of building abstract, pluggable interfaces that go into an abstract enterprise service bus and creating the right sets of interfaces and managing them needs to be driven by a methodology from Rational, and by people like you, Grady.
GB: How do you see the larger Rational mission to promote iterative development practice, and what challenges do you see ahead?
DS: There are many challenges. One is to get people to understand that iterative development is extremely important, for several reasons. The obvious reason is that iterative techniques help reduce mistakes, with more checkpoints along the way to help teams understand exactly what's going on, as opposed to the long running waterfall processes that were in place before.
But just as important, many of our customers in today's corporate spending environment are looking for a more incremental approach to improved functionality. They need to see return on investment quickly, which means they're more willing to invest in an iterative approach that maps to an incremental, business-oriented model. This is what we're striving to establish within the Rational brand -- a strong tie between incremental refinement in business goals and iterative, RUP-oriented development methods. This will quickly lead our customers to a tighter alignment of tool investments with their business goals.
GB: To what extent are Rational tools being used by IBM Global Services (IGS) in their customer engagements?
DS: As IGS moves to a more asset-based model, away from a hardware and labor-based model, the Rational tools and methodology can be a large part of IGS's success. IGS is discovering an ever-increasing reliance on software development -- the creation and management of software assets, and an understanding of the application lifecycle that requires reuse of those assets. These are defining elements of IGS's transition, and it goes to the heart of what we're trying to accomplish with Rational tools within IBM Software Group.
There's a growing partnership between IGS and SWG -- an important part of which is the relationship with Rational. IGS will become more successful as we help them build skills, methodology, and systemic execution around building and managing assets, all of which is important to the evolution of their business model.
GB: When leaders change, there are usually some areas of the business that receive special focus. What does your first 100 days look like?
DS: We're going to aggressively work on several things.
- Making sure that Rational tools are more tightly connected to our customers' business goals by closely focusing on our customers' actual runtime environments. This means taking the tools we currently have and orienting them to the types of runtime environments in today's businesses. We do a great job of dealing with modeling and methodology, but we need to take our development environments and our testing environments and our performance management environments and better integrate them with the realtime runtimes that our customers use to drive their businesses. For example, dealing with connections back into off-the-shelf or packaged applications, and extending that to other packaged applications and legacy applications. We need to pursue better integration with the assets we already have within SWG and other divisions, and bring those capabilities and successes into Rational.
- We're going to continue the work started with Al Zollar's team at Tivoli to enhance the integrations with system management and operations management tools. We're trying to align the assets created at development time, and help bridge the gap between operations and production/development. In a lot of custom development and systems integration shops, you find that developers and operations people are at war, because developers work in tooling environments designed around simplistic assumptions. Not that the tools themselves are simple. But the environment in which applications are developed typically is very different from the environment where they get deployed -- and scaled up. What we're trying to do with Tivoli is to ensure a greater understanding of the challenges to all the teams involved -- the developers, operations people, and the systems administrators.
- We're going to integrate further with Eclipse. This effort will involve our data architecture tools, as well as business process modeling tools, business process management tools, plus portal creation and management tools. We're going to do this in conjunction with the WebSphere and Lotus brands, and extend this out to the Tivoli brand. Essentially, we're going to do a lot more integration with the individual SWG brands, mindful that the Rational's value reaches beyond IBM's own software offering (as is the case with every other SWG brand). So we are going to continue to focus on functionality in heterogeneous shops, because that's the real environment our customers work in.
- We're going to be paying close attention to the evolving collaboration with the community of software developers. Rational's tools need to adopt not only to lifecycle management for application development teams, but also lifecycle management for a notion that's building up around the Internet, of communities. Look at the dynamics of geographically distributed development. Everybody wants to leverage community collaboration capabilities available across the Internet, whether that's blogs, RSS feeds, instant messaging, portals, or geographically distributed source code library tools. All of these things are already being leveraged by the open source community, and they need to be part of Rational's value proposition. We want to embrace that notion of collaborative community, as defined by the Internet, to the evolution of development teams working on asset management.
Not that this is radically new. But the Internet introduces new capabilities and greater reach, which enables things like globally distributed development, helping you decide how and when to outsource or insource, and offering you more flexibility in managing development assets -- including intellectual property and the people who create it. Our customers will demand these capabilities as their needs become more global.
GB: Rational's legacy management team has now largely been replaced by IBMers. Will Rational undergo a change in leadership style over the coming years?
DS: The IBM management profile is really no different from the people who have been at the helm of Rational itself. I've known Mike Devlin since the time Rational was founded (which means I've been in and around Rational since the old Ada days). Now, it is true that we're bringing Rational forward within the much broader context of SWG, but that's the primary reason Mike thought the Rational acquisition was a great idea in the first place. He was concerned about his abilities to scale his business without partnering with a larger company like IBM.
The basic Rational mission is not going to change. The number one objective is still to ensure the success of organizations that depend on developing and deploying software. We'll keep finding ways to do that, whatever it takes. In Rational's early days, that meant designing and building original hardware to run an Ada development environment. That's consistent with my own level of commitment. Most people don't realize this, but when I first came into the IBM Research division, out of graduate school, I actually brought in Symbolics LISP machines because it was the best hardware at the time for what we were trying to accomplish.
On a deeper level, something that's contributed most to Rational's success over the years is working closely with customers. And that's something I've long been known for. For the past seven years, I've averaged about 150 partner and customer contacts per year. I really like doing this, because it gives me a very connected sense of how IBM's business value is benefiting our customers. In some cases, it shows me where we're failing to deliver on that value. I get a very down-to-earth understanding by doing all types of customer calls, including the ones where I find out what we're doing wrong. I intend on doing as much customer contact as possible, and I will infuse it into the culture of Rational even more than it has been in the past.
GB: What inspires you most about leading the software development brand at IBM Software Group?
DS: For many reasons, IBM Software is important to IBM's success. The most profitable portion of IBM continues to be software. And as far as Rational is concerned, there is tremendous opportunity to extend the value that Rational has already established over the past 25 years. It's pretty clear that developers are becoming more and more important to the success of an evolving business, that they need to be more tightly connected to, and understand, the value that they bring to businesses. It's high time that CIOs, software developers, and systems integrators received an opportunity to become more tightly connected to the evolution of business models and business returns. I think this is where Rational can bring value to the marketplace and extend its current value proposition -- of not only making developers effective, but also making businesses effective in the ways they relate to their software developers.
About the author | | | Grady Booch is recognized internationally for his innovative work on software architecture, modeling, and software engineering processes. All of his work has improved the effectiveness of developers worldwide. Grady is one of the original developers of the Unified Modeling Language (UML) and was also one of the original developers of several of Rational's products including Rational Rose, the industry leading visual development tool. Grady is the author of six best-selling books, has published several hundred technical articles on software engineering, and has lectured and consulted worldwide. Grady served as Chief Scientist for Rational Software. |
Rate this page
| |
