IBM iron still seems to be the preferred choice over client-server farms for some businesses with huge volumes of information to crunch fast.
It was what 407 Express Toll Route (ETR) – the operator and manager of Highway 407 – opted for when the highway’s management switched from the public to private ownership by a consortium of companies in 1999. Highway 407 extends 108 kilometers across north Toronto.
407 ETR is sometimes cited as a model for cities struggling with issues around traffic congestion and urban sprawl. Traffic congestion in the GTA costs almost $2 billion per year in lost time and productivity. 407 ETR seeks to relieve congestion on other major 400 series highways. It serves 320,000 customers on weekdays.
Instead of physical toll booths that cause delays and line-ups, 407 ETR relies on technology to bill motorists who use the highway.
Transponders are used to log entry to and exit from the 407, and are available for purchase or lease by customers. Overhead sensors log access and bill customers on a monthly basis. For customers who don’t have transponders, license plate recognition systems are used instead. The system generates millions of micro-payment transactions daily.
All systems, from roadside to back office billing, have been systematically replaced and upgraded since the company was privatized, says Dave Allred, CIO of 407 ETR.
The company installed its first mainframe in January 2000, and it has upgraded twice since, purchasing a new box in 2003.
407 ETR uses its mainframe as a client-server system. The mainframe is the database server and manages all data, and Web-based applications access that data. The system is effectively a thin client server environment.
Allred says they considered both mainframe and client server systems at the outset. “When we did an RFP to look at server farms needed for our requirements, we looked at both thin and thick clients,” he says. “We settled on thin because we wanted to be able to run applications anywhere from people’s homes. We wanted to eventually be able to push those applications to customers.”
Several key features drew Allred to a mainframe-based system. The mainframe’s legendary reliability was a top selling point. “Since December 2003, we’ve had no unscheduled downtime whatsoever,” he says. Security and data integrity features were also key. “It’s difficult to hack into, and protection of data is easier to manage on a mainframe.” The throughput per engine was also a big draw. “We needed the speed of each engine versus pure horsepower. We needed each engine to be as speedy as possible.”
These features were all attractive, Allred said, but the final determining factor was the ease with which a mainframe can be managed. “It came down to scalability, reliability and the maintenance costs from a people perspective. This is an extremely [big] selling point. We have a small staff that can process an awful lot of transactions on a mainframe.”
Such people factors are becoming increasingly important to many businesses. “At the end of the day, a mainframe is very easy to maintain due to centralization, compared to a server farm with 40 or 50 machines.” Allred described the scales used to evaluate the two. “We would have to trade off salaries against boxes, security, reliability, power…all of the above.”
George Goodall, an analyst with London-based Info-Tech, points out that this becoming an increasingly critical issue. The analyst firm has dealt with clients who have inherited a lot of servers through consolidation, acquisition, organic growth and so on.
Their feedback, said Goodall, was summed up in one individual’s comment. “It’s not the cost of the servers that bothers me. It’s the two full-time equivalents that go with each server.’ That is the real cost driver.”