Wayne Fowler: Open Systems Critic and Champion

He helps Toronto Stock Exchange take a multivendor path

Biography

            Name: Wayne Fowler
             Age: 48
  Place of Birth: Chatham, Ontario
Current Position: Director of Technology Network Services, 
                       Toronto Stock Exchange.
   Years in Current Position:  3
       Years in the Industry: 24
               Car He Drives: 1986 Toyota Tercel
Favorite Non-work Activities: Toy making, crafts, soccer coaching,
                                  science fiction reading.
      Pet Open Systems Peeve: "Open systems are a myth. Vendors 
selling into the open systems marketplace have never gotten their 
act together. The concept is very elegant and very doable, but I
think open systems vendors in general are in retreat and I don't
believe that ever should have happened."

Putting open systems into practice has been Wayne Fowler's preoccupation for six years. So it's a little strange that Fowler doesn't believe that open systems are real. "There really are no such things as open systems," he says, letting the statement hang for its ironic effect. Then he explains: "Open systems vendors in general are in retreat. They're in retreat from things like the desktop, and from the bottom-end server. And they're not taking over the high-end mainframe marketplace as quickly as they should." Now the statement makes sense; for him, open systems should not be a sometime thing. He wants a perfect world.

Fowler has been in a position to evaluate the situation as he has helped convert the Toronto Stock Exchange from a mainframe-only environment to one that includes 40 UNIX servers and 250 networked UNIX desktops. He believes open systems vendors had chances to push ahead but didn't. "Typically, they're not winning where they should be winning, and I think that's a shame," he says. "I think the opportunity was there to do that, but it wasn't taken up." As a result, the promise of having open systems across a broad scale of equipment has receded. "I think all the major vendors are walking away from the desktop and furiously trying to get into servers. Industry is in retreat from being able to run open systems from the desktop to the scientific, massively parallel computer. It's unfortunate, because there was a window of opportunity there-some of which is still there-but didn't get taken advantage of."

A Native Canadian

Fowler was born and raised in Chatham, in southern Ontario, about 40 miles from Detroit. While a senior in high school, his math teacher took the class for a weekend visit to the University of Waterloo that included a chance to play with the university's computers. For Fowler, that was the deciding factor. "If I was going to do anything, it would be in math and computers, and Waterloo was the only place where I was going to have first-year courses that would actually let me do programming," he says. "Waterloo has had a long-standing reputation in computer science and engineering and turns out several hundred high-quality computer science grads every year."

Fowler earned his bachelor's and master's degrees in computer science at the university and then worked there for four years. The business climate was a little slow in 1971 and the university was looking for systems programmers. "I didn't know what they were but I took the job anyway," he says. Working on IBM 360-class machines, Fowler did technical support and spent time testing system changes, writing a text editor for the then-new VM operating system, and looking after an intelligent, programmable telecommunications front end. "It was fun," he says. "I got access to all the really interesting technology."

By 1975 it was apparent that the university's systems were becoming obsolete, and the funding to keep both equipment and staff going was drying up. Fowler decided to take a job with the Bank of Montreal. As a part of the bank's systems programming staff, Fowler had a role in launching all-night branches, real-time updating of accounts, and completely automated operations. "It was the first bank in Canada with real-time updates," he says. "They started converting all their branches from manual to completely automated in 1975, and it took about six years to convert 1,200 or 1,300 branches in five time zones." Over a 12-year time span, Fowler's responsibilities changed from mainframe-only to technical support for every kind of system but mainframes. "I started to get more exposure to computing with something other than this $5 million piece of iron," he says. "I started to get familiar with these odd little companies like Digital and Tandem." After nine years in a multivendor environment, Fowler found himself managing 40 employees and looking after 11 data centers in Canada, the United Kingdom, and the United States, and tending everything from disk drives to small mainframes.

End-User Concerns

Fowler's activity in end-user groups also began at that time, with his involvement in a systems management requirements group for IBM users called GUIDE. Between 1979 and the mid-80s, users were starting to face up to the large systems management issues. "Companies like Arco and American Airlines were involved in a lot of projects to get IBM to better manage large-scale systems," he said. Fowler published papers on security and on storage management, and also earned the Bank of Montreal's President's Award for his work in "helping IBM see the light in terms of listening to their customers and getting agreement on how to manage large amounts of storage."

Beginning in 1984, Fowler set up and managed end-user technology for the bank's employees, who were beginning to work on relational databases and use electronic mail on PCs. Fowler had to personally approve the purchase of 3,000 PCs, including setting corporate standards and managing acquisitions. "It was a fun time," he remembers. "We spent a lot of time thinking about alternatives to large mainframes, and that's when I first ran into Oracle." It was also his first encounter with the open systems idea. "We started to get into the concept of applications that were independent of vendors," he says. "Oracle was running on DEC VMS and Data General equipment, and you could actually make a hardware decision without having to worry about whether the application would run. It was the first time we actually got into the concepts of interoperability and commodity hardware, in the case of PCs, and whether it did or didn't exist." Although he was familiar with the idea of a generic operating system from his university days, the presence of database portability and multivendor PC hardware made it real. "All those concepts really happened when I got warped and twisted in the end-user computing environment, and understood that you didn't have to have a large-scale IBM mainframe to actually do the job for people," he notes.

The Move to Toronto

Fowler moved to the Toronto Stock Exchange in 1988, where he has remained, taking on increased responsibility over the years. The Toronto exchange, the third largest stock exchange in North America after the New York Stock Exchange and NASDAQ, was looking toward building a new set of systems, and also wanted its systems to support new trading services. The alternatives in 1988 were considered to IBM, Tandem, or a possible third choice. Fowler got the responsibility for devising corporate standards for the systems, so he worked out an overall architecture that would take the business into the future. It turned out to be a complete UNIX environment connected with TCP/IP protocol.

Why go with UNIX? "We needed to have the business opportunity and the appropriate level of technology to satisfy it," Fowler says. "The first opportunity was that we needed to explore how we would build a new trading engine. Early on, we had chosen Oracle as our corporate database standard and were running it on our mainframe. Oracle, at that time, was going through a fairly significant transition from being a Digital database supplier to seeing lots of action going on in the UNIX world. We were following that growth, and we were seeing a lot of very powerful commodity hardware coming from people like Pyramid, Data General, Sun, and HP. So we were starting to see people move into the UNIX marketplace with viable products."

So in 1989, the exchange had decided to provide a new trading service, and to build it on Oracle software. They asked for hardware bids from both UNIX and proprietary vendors. The only requirement was that it run Oracle. They chose Data General Aviion UNIX-based systems for the new service, and Pyramid UNIX symmetric multiprocessing servers to be a prototype for the new trading engine. The new service was in production three to four months later. "We were able to actually plug it in, get it running, and be able to afford a backup [system] sitting there-all these kinds of things that people say were supposed to happen when you get open systems-a choice of hardware, competitive bidding, portable applications and all that stuff. We said 'Hey, this stuff kind of works.'"

A second development came in early 1990, when the exchange faced the problem of providing improved trader access to the trading floor. The choices were considered to be a UNIX desktop running on proprietary hardware, SCO UNIX running on PC hardware, or PCs running either OS/2 or Windows. "We said that, given a choice, a safe bet is the SCO Open Desktop running on PCs, because if we really blow it, at least we don't have to buy the hardware again-we just go over to the other two choices." Also, the exchange needed to build most of its own applications, so it needed a rich development environment. UNIX was considered to be reaching that point quickly. So the Toronto exchange became one of the first businesses in Canada to use SCO Open Desktop on IBM PCs, with about 150 SCO-based machines supporting its floor trading.

TCP/IP Seemed to Fit

The choice of TCP/IP protocol seemed to fit logically with the other choices. A client/server architecture with message-based structures had been one of the first decisions. After that, "We were finding more and more that the open systems marketplace was supplying us what we wanted in that environment," Fowler says. For example, when they had to connect the SCO Open Desktop applications to the Tandem system, it was simpler to give the Tandem system some LAN ports and plug in TCP/IP than to make the PCs talk to the Tandem system via proprietary protocol. "And as we went looking for products, we were finding them in the open systems marketplace," Fowler reports. "We found TCP/IP to be a good, universally supported choice for networking. So as we were moving toward implementing the architecture, we were implementing it with open systems choices, not necessarily at the expense of replacing every proprietary system in the shop, but making those proprietary systems coexist."

One interesting side effect of the emergence of open systems choices has been on the evolution of proprietary systems, Fowler likes to point out. The prices of proprietary systems have been slashed to compete more closely with UNIX servers, and more options are available that make them connect with open systems networks. "The proprietary systems we have are becoming more attractive," he says. "A classic example would be Tandem. If you went to Tandem six years ago and said you wanted to buy a processor, they would have sold you a processor for $4 million. If you go today, you get an equivalent processor for less than $1 million because they're using commodity hardware. And you can buy a processor, plug TCP/IP native on it, and it will even run Tuxedo [UNIX-based transaction processing software]. So the proprietary vendors that are offering UNIX systems as well, are applying what they have learned to their proprietary systems base. So now it's a much closer comparison for the buyer. You've got to be a much better tap dancer these days with your open systems choices."

Today, the Toronto Stock Exchange is three months into a pilot project to bring traders even further into the information revolution by eliminating the trading floor itself. Using a UNIX desktop, the trading of equities would be completely automated and trades would be executed electronically without the yelling that takes place on the floor. The network for the new system is managed with HP systems running OpenView for network management and customized code. "That's something we are now fine tuning," Fowler says. A number of its member firms already have direct links and do message-based access into the exchange's systems. The Singapore and Paris stock exchanges are taking similar steps. "A lot of people are watching us, and a lot of people are concerned about how long it will take us to do this and how much we're spending."

The move to a floorless exchange also has emotional connotations. "There's a lot of long-term investment in people's lives being a floor trader," he says. "It's a pretty tough thing for an exchange to do, because, ultimately, those floor traders work for the guys that own us. So, independent of the technology, just the fact that we're doing it is a tough thing. We're about 150 years old and we've always had people yelling at each other when they do trades. And what are you going to show on the evening news as background when you don't have people yelling at each other on a trading floor anymore?"