McKinley Rides Merrill Lynch's Second Tech Wave

By Maria Trombly | Securities Industry News | June/July 2002

John McKinley, Merrill Lynch's chief technology executive, saw his first major tech downturn when he was still in grade school. It was back before the era of personal computers, when he was 10 years old, and his school had received a computing grant from the National Science Foundation. He got hooked on computers-and then, to his chagrin, the grant ran out.

To feed his new appetite, he started to "borrow" computer time from universities.

"I loved it," he recalled. It was back in the mid-'70s and McKinley discovered the ARPA Net-an early precursor of the Internet. "Those were amazing times," he said.

But despite his early self-proclaimed "rogue" background, McKinley wasn't destined for a life in the dark shadows of the fast-growing computer world. Instead, he went to college, majored in business and went to work for Atlantic Richfield Co. (now part of British Petroleum) right after college.

"They had a fantastic boot camp," he said. And his experience there taught him his way around the corporate computing environment, even allowing him to play with new technology coming out of the Xerox Palo Alto Research Center (Parc)-the center that developed the first graphical user interface, for example, which was popularized in the Apple Macintosh and, later, in Microsoft Windows.

One of his major projects at Arco was helping to build a computer center for business users. "There was a tremendous appetite for analytical tools," he said. "Unfortunately, the company was structured in such a way that users had to request reports and wait-and wait-before getting results."

To eliminate these long cycle times, McKinley helped build a decision-support center-a precursor to today's data warehouses-which, in effect, helped open up the company's computer infrastructure to a larger community of users. After that came a 13-year stint at Ernst & Young, where he worked as a consultant focusing primarily on financial services, and where he helped build a software development center at the firm, collaborated with IBM on a major outsourcing project and helped run IT centers in client corporate organizations.

In 1995, he left Ernst & Young to become the CTO of GE Capital, where he worked with recently retired GE CEO Jack Welch and the CEO of GE Capital Gary Wendt (who now runs Conseco). Then, at the end of 1998, he came to Merrill Lynch, which had a reputation for "not getting it" when it came to the Internet. Technology-savvy upstarts like Charles Schwab and E-Trade were threatening to take away business by lowering commissions and offering self-service Internet-based trading. Firms like Merrill Lynch, committed to large staffs of commissioned brokers, started to look like dinosaurs.

"We were being painted as the before' picture of e-commerce," McKinley admits. It's an easy admission because under his technology leadership, Merrill Lynch transformed itself from a dinosaur into a technology leader.

"We got the wake-up call and responded with a full-court press on every front," he said.

Merrill Lynch became the classic "second-wave" success story. While the first wave of companies jumped on the leading edge of the Internet, and often crashed, Merrill applied the emerging network technologies strategically to its business.

"The industry has a mixed record using the Web as a client-acquisition channel," McKinley said. "A more successful approach was to use it to offer always-available global services."

Using Internet technologies to permeate the business at all levels would help the company cut costs, improve service and raise quality. For example, looking forward to the needs of next-day settlement and straight-through processing, Merrill Lynch began involving clients in developing and maintaining their own data using the Internet.

"You have to build in quality at the time the data is created," he said. "The best thing in the world is to get the client involved in the qualitization of their information. Then you want the client actively involved in the exception process."

Similarly, networking has become a critical part of almost every business process at Merrill Lynch. Both the public Internet and the company's own Intranet are now being used to provide telephone service, to provide employees with a unified view of their customers, to streamline communications among employees and between brokers and their clients, and for a number of other uses that connect people to one another and to computer systems.

This technology has helped the company recover from the effects of Sept. 11, when Merrill Lynch had to evacuate its headquarters near the World Trade Center and move across the river to its New Jersey facilities. Wireless networks, telephones that use Internet protocols and Web-enabled data centers and servers all helped the company get back on its feet quickly.

Recognizing his contribution to the company, Merrill Lynch promoted him to executive vice president last year. In addition to being the chief technology officer, he is also head of operations, corporate services and security-and is well-positioned to drive the second phase of Merrill Lynch's technological transformation.

This year, McKinley will start to turn the entire computer infrastructure of the corporation into one single brain using an emerging XML technology called Web services.

Historically, the problem with integrating corporate systems at Merrill Lynch-and at any company-has been that applications are written in different languages, run on different machines, store data in different quirky formats and interact with users (and one another) using proprietary protocols and interfaces. An entire industry has arisen to help solve integration problems, feeding whole armies of consultants.

But after the success of the Internet-and the universal adoption of HTML as a common format in which to present information-users began wondering if it was possible to create a common format in which computer applications could exchange data.

Over the last two years, as XML has finally grown up and hit prime time with a set of public protocols that describe how applications find each other and interact with one another, Web services will become the standard to which applications are written.

"Web services is the biggest thing we're emphasizing architecturally this year," McKinley said.

When new applications are written, they will be Web services from the start, he said. They will speak XML, Soap and WDSL, and this will increase speed of development, testing, deployment and reusability.

The problem, McKinley said, is that it's impossible to go back and rewrite all existing applications and it's too expensive to replace them all over night. And many have worked fine for decades and could conceivably work fine for years to come.

To solve this problem, McKinley said, as a transitional measure, legacy applications will be wrapped in a shell that exposes them as Web services.

From the outside, they will look like shiny, new applications-programmers will be able to access them through easy-to-use development tools like Visual Studio .Net without having to know the internals of older technologies like CICS. Inside the shell, the old applications will continue to function the way they always have, without having to be aware of the fact that the world outside has completely changed.

Each shell will have to be customized for the application it serves; it will need to understand the way it stores data and communicates with users in order to translate everything into common standards. But it will be easier to create this shell than to rewrite the application from scratch.

"I'm going to make a series of incremental steps," he said. "Replacing bespoke systems with commercial packages, replacing mainframes with distributed systems."

Eventually, the applications inside the Web services shells can also be replaced. As long as the Web services interface remains consistent, users and other computer systems will never have to know that the mainframe applications have been replaced with one running on, say, Linux.

Moving to Web services will mean that Merrill Lynch can switch vendors at will, or mix-and-match applications from providers to choose the best performance for the lowest price. That's bound to create headaches-and resistance.

"Every technology provider we have is operating in a pretty stressful environment," said McKinley. "Some vendors are trying to corner the market on Web services by developing their own, proprietary versions-and trying to lock customers into their visions of the future."

But ironically, the same economic downturn that has hurt Merrill Lynch's coffers is helping it make its case for open standards with its vendors.

"It's basically a buyer's market," said McKinley. "We are in a broad-based technology deflation. That allows me to have a significant leverage in what I want to see. As buyers of technology, we don't have to have five different answers to Web services. We won't accept the divergence path that the vendors want us to go down."

Further, he said: "I want development tools that can support Web services regardless of their targeted end-state platform. I don't want to have different answers in security, discovery and Web brokering."