The death of mainframes actually was supposed to occur in the 1990s. Technology writer Stewart Alsop, the former editor of InfoWorld, predicted in 1991 the last of the mainframe computers would be un-plugged in 1996. That never happened, though, and likely never will. New mainframes are being built and sold today, particularly by IBM. Those who predicted the death of mainframes were misguided, Jamie Bisker, director of research for the insurance practice at TowerGroup, believes. The demise of the mainframe was a logical prediction made by people who were inadequately informed to make those decisions, he maintains.

Similarly, Paul DeFuria, chief technology officer for CSC Financial Services Group, points out the mantra for insurance IT leaders today is leveraging existing IT investments, not eliminating them. Organizations almost always will gain durable competitive advantages by exploiting new features and capabilities of electronic platforms over any replacement technology. Nothing on the horizon is good enough [to convince insurers] to start over.

Wrong Guess

The prognosticators who foresaw the demise of mainframes were some of the same people who said insurance agents were going to vanish and insurance will be purchased almost exclusively on the Internet, according to Bisker (he wasnt among them). Technically, theres no reason mainframes couldnt [disappear], but pragmatically, its too complex, he says. Just as the advent of the telephone did not get rid of paper mail, computerization and the Web will not get rid of everything else.

And whats not to love about the mainframe? Maynard points to two important factorsreliability and scalabilitythat make the mainframe so valuable. All the things we know and love about the mainframe are things valued by most big insurance companies, he says. Many of the large companies run billions of dollars through their mainframe computers.

Security issues on the mainframe are rarely a problem, something that cant always be said outside the mainframe environment, Maynard believes. We have to work hard in the middle-tier environment to get the equivalent of those kinds of protections, he notes.

Jim Lupton, vice president in information systems with American Fidelity Assurance Company, agrees. People are realizing it costs a great deal of money to move working, stable mainframe applications to network systems for the purpose of accessing data in a distributed processing environment, he says. The bottom line is: The mainframe is a mature server platform. Its powerful, more stable, less expensive overall, and supports a tremendous number of applications in one place.

Mainframes bring the concept of nines into focus, adds Bisker. What level of reliability do you want? he asks. Five nines99.999equates to five minutes of downtime a year. Mainframes can achieve that. There are other platforms approaching that level, and some may have achieved it, but they are not the rule of the day.

No Flexibility

Reliability makes you think of the family sedan. Flexibility brings to mind a sleek sports cardemonstrating amazing cornering capability as it accelerates through the turns. Maynard says The Hartford uses mid-tier platforms along with mainframes, and the platforms are much more flexible than the mainframes. Were able to bring to market a lot of computing power for the price [with mid-tiers, more] than we are at the mainframe.

Mainframes are constrained in what they can do, and as Maynard indicates, You can order a number of servers and be up and running in a short period of time. With a mainframe, you cant do that.

What The Hartford and other carriers must do is to try to find ways to make up for that lack of flexibility. In our particular case, were going through a renovation and consolidation effort on our personal lines systems, reports Maynard. A new policy administration system was purchased. It exposes a number of the back-end services via a Java middle layer with ACORD XML, which is a standards direction in which we want to go. Its a combination of the middle-tier software technology and the mainframe reengineering, if you will, the standard products, he says.

Newer mainframes do offer some flexibility from an equipment and operating system standpoint, Bisker contends. [Newer mainframes] have tons of system engineering tools to keep them running. They are fault tolerant, resilient, and all those other nice words we like to hear, he says.

Legacy

Mainframes often get coupled with the word legacy, Bisker maintains, and that is one reason for the negative publicity theyve attracted over the years. Its not the mainframe itself that is the primary element of what we term as the legacy environment, [mainframes] just happen to be the equipment the legacy systems run on, he says. Certainly, if the code that ran on the mainframes kept pace with the technical improvements that have been happening with mainframes, we wouldnt have a legacy problem. The way I classify a legacy system is its not the age or the language its written in, its how [the systems] are designed.

Legacy systems are not designed for ease of use or ease of change, Bisker continues. They tend to be proprietary in the fullest sense of that word, meaning the system was designed to execute on a given platform in a given way without much thinking of how things might change in the future, he says. You could have written a perfectly flexible, adaptive system in COBOL, and you still could do it today. Its old, but it still works.

Maynard agrees. The policy administration system purchased by The Hartford utilizes CICS, DB2, and COBOL. Those are pretty standard 60s and 70s kinds of technologies, he says. But theyve matured over the course of time. The mainframe hardware and software technology are getting better, as well, to make them more flexible for us to take advantage of new features.

Adding Functionality

Having access to the data on the mainframe is a critical issue in the mainframe environment, Bisker asserts. You dont want to rewrite or reengineer a system, he says. You can do the wrappers with Web services or any number of middleware-type solutions.

That is what American Fidelity accomplished. The carrier needed a Web portal for its agents, and Lupton says he had no doubt such a connection could be achieved. We either were going to have to rearchitect to change the way some of the systems work, put out an extract, or we could go real time against the data using the Enterprise Transaction Systems from Software AG, he says. We put together a real-time engine that allowed us to go directly against our systems. All the security is built into the main system, so the users do exactly what they did before, and it automatically enables the Web for the individual customer.

The key to successful platforms is having good software to operate on the systems, according to Bisker. He doesnt believe there is as much software being developed by third-party vendors for mainframes today but adds the primary competition to purchased software is homegrown software. In this day and age, companies are looking at the buy- and-build model, he says. They have this capability [to build], so they make use of it and get more work done.

Staffing Needs

The Hartfords personal lines company is installing more than a new policy administration system, Maynard reports. There also will be a new rules engine and rating engine. Because were going through these changes, we have to figure out how were going to position the staff to take advantage of some of the newer technology, he says. This is fairly standard technology, but it is a different mindset for us. In prior years, The Hartford most likely would have developed and built everything in-house. That has changed, though. Has our internal mix changed? he asks. No. Is it going to change? Yes, we definitely will have to retool some of our individuals as we move into the newer implementations. We are putting the changes of the applications in the hands of some of the business folks because weve externalized and table-ized a lot of things to allow [the business units] to make the changes.

There is resurgence within the industry of having good IT shops, Bisker states. IT managers are asking themselves, Are we good at doing what we think we are good at? If we arent, maybe we should get out of that business, he says.

A significant portion of an IT departments budget goes to the maintenance of existing systems, Bisker adds. That is one of the real challenges, he asserts. [Legacy systems] represent a maintenance dilemma. No matter how efficient or wonderful they are, they do need to be maintained, especially if there are enhancements that are needed. Do we want to be in the maintenance business, or do we want to use our IT capability for new systems, integration, analyticsthe things that are necessary today to succeed in business, particularly insurance?
DeFuria also believes the IT team can make a difference. You have the best operating system in the world, he says. Its the most sophisticated and has the most capabilities, but it is not trivial. The ongoing challenge [insurers] face is they need a team that is good. To be an IBM systems programmer definitely is an advanced skill.

If a carrier has moved away from mainframes, it may face a whole new set of problems, Paul Manning, lead architect in the property/casualty division at CSC, points out. There are a lot of cases where you have IT departments being staffed by younger and inexperienced people, he says. By inexperienced, I mean they havent worked with an enterprise-sized system. They recommend throwing a server in and giving the appearance of a high-end system, but then the system starts to scale, and now you dont have one server, you have a hundred servers strung together with miles and miles of cabeling and no way to manage the deployment. Companies have gone from tens to literally hundreds of servers, and they have no idea how to manage them in anything approaching a cost-effective way.

The infrastructure team with The Hartford personal lines deserves a great deal of credit for maintaining the old and bringing in some new features, Maynard declares. It has done a good job from a mainframe and mid-tier technology perspective in helping us to keep up without breaking the bank. We make sure we are spending the money prudently for our business, he says. We need to make sure were taking care of those assets and not spending the money foolishly.