Does high-availability warrant the high cost?

The old adage “you get what you pay for” need not be the blueprint for boosting reliability and high-availability in the financial services sector. Loren Hicks, a long-time banking and securities consultant and the Managing Director of Toronto-based Visioneering Partners Inc., likens the quest for high availability – that is, being available for your customers whenever they need you – to picking a point on the curve created by the percentage of time the service is accessible and the cost of achieving that objective.

And while an initial outlay of, say, $2 million, can create system availability somewhere in the neighborhood of 99 per cent or better, the costs for a negligible increase can quickly spiral out of sight. Therefore, if availability were represented vertically and cost horizontally, the line to achieve the benchmark of 99 per cent availability would be nearly vertical. However, ensuing improvements to reach 99.5 per cent, for example, would be represented by a nearly horizontal line.

The question, says Hicks, is where companies in the financial services industry want to position themselves on this curve.


Historically, Canada’s big banks have positioned their branch systems at the extreme edge. They offer customers unparalleled availability and reliability through proprietary networks and legacy systems consisting of rooms full of huge mainframe computers running in lockstep with identical systems hundreds or even thousands of miles away.

These systems were designed to serve tellers (and later, ABMs) and process the same data. If the system failed in Montreal, for example, transactions could simply be re-routed to a different server – thereby avoiding costly downtime and, more importantly, customer frustration. If one teller’s machine goes down, the next one is identical, and if a branch is down, the branch or ABM just down the street has the same systems and immediate access to the same data.

The downside to such high-availability is the huge expense of such proprietary networks that comprise redundant equipment, duplicate organizations and a large human resource and policy infrastructure that manages it all.

These systems are required to operate with virtually zero defects, says Hicks, and the banking IT bureaucracy dictates that even the most minor changes be debated at length. If the changes are approved, they are frequently subjected to a battery of tests before a costly implementation that could involve re-training thousands of tellers. The philosophy and culture of these organizations is to avoid risk at all costs.

The costs for the upkeep, maintenance and possible enhancements to these systems are considerable, says Hicks, and comprise a sizeable portion of the transaction fees paid by customers. Still, the fact that Canada has a few major national banks operating thousands of branches spread across the country in six different time zones also makes fail-safe systems such as these an absolute necessity.

“There’s no question that it’s still the right thing to do for branch banking,” says Hicks of these massive systems, “because of its reliability, the number of devices out there and the fact that you want a homogenous environment. You want to know that the teller in Vancouver is running the same software as the teller in Halifax, otherwise you run into training problems, human resources problems and procedural problems. It’s got to be identical.”

As the emergence of Internet banking has given increasingly fickle customers an unprecedented number of choices in the financial services industry, it’s imperative that banks keep them satisfied. “(Banks) want reliability like nobody else does,” adds Hicks. “If they’ve got a problem with software that does something ‘funny’ with a loan, then something ‘funny’ has happened to five million loans across the country, leading to five million angry customers.”

Hicks cites an incident in the 1970s when a branch system was accidentally run overnight with the wrong date, thereby affecting millions of accounts. Daily interest was calculated incorrectly, mortgage payments were taken from accounts before paychecks went in, and so on. It took any army of clerks several months to clean it up, millions of dollars in extra costs for the organization, and an unknown toll on client confidence and satisfaction.


While the use of such systems is justified in inter-branch banking, they were never the right strategy for other financial service providers. “Even when the biggest Canadian banks were among the top 20 institutions in the world during the 1970s, their non-retail operations were relatively small in global terms,” claims Hicks. “Now that even the biggest Canadian bank isn’t among the top 50, the best that any can hope for is to be a big player in a very few niche markets.”

For example, securities businesses operated for many years with cheaper systems, as they were unwilling to spend the extra money and lead times to climb further up the availability curve. Spending millions of dollars on systems improvements was unnecessary, says Hicks, since the phone was an adequate backup in all but the most severe outages. Property and casualty insurance businesses were able to keep costs down because they had few real-time client requirements prior to this decade. In fact, much of today’s domestic life insurance businesses remain well supported largely by batch systems.

However, the Internet has transformed much of the financial services into a self-service industry, and different businesses now have to decide where they can afford to be on the availability curve. “Super high-availability is the wrong thing to do for a lot of e-commerce,” Hicks says. “It is not yet cost effective, and customers don’t really need it for a lot of their business, but there are strategies for high-availability over the Internet that are inexpensive and reliable.”

In fact, Canadian financial institutions are already facing pressure from online financial services that may boast slightly less reliability, but charge customers considerably less in transaction fees. For example, a small business with an Internet presence may make the sacrifice in system accessibility if it means paying only 1.5 per cent per credit card transaction to an online service, versus over three per cent to a big bank because of what Hicks calls a “presumed higher fraud rate.”

“The new entrants are counting on Canadian banks being unable to lower fees,” says Hicks. That, and access to the payments system.

The Canadian Payments System was closed to all but banks and credit unions until the 1990s, when pressure from securities dealers and insurance companies seeking the ability to offer deposit accounts with ABM access, etc. led to deregulation. The payments system remains closed to many financial services firms, preventing already low-fee services currently operating in the United States from coming to Canada, Should the climate ease, however, a small player with a couple of Unix servers, a good Internet front end and a back end linked to credit card and debit processing could offer e-commerce services to small businesses at a fraction of its current cost.

Small businesses will get to choose how they handle occasional outages, says Hicks, by either living through them or by having two providers and switching between them. Redundancy can be attained less expensively this way than by having one provider move to 99.9 per cent on the availability curve. The small business can get to almost 100 per cent availability by having two service providers each at 99 per cent. The securities industry has been doing this for years with the market information and quote vendors.

The choice between availability and cost also applies to insurance, says Hicks. “Say you are at only 99 per cent availability. If someone gets bumped from your company Web site and never comes back, at most you will have lost only one per cent of your potential business. Multiply that by your close rate and your profit per policy to find how much you have lost. It will almost certainly be a lot less than the costs of super-high-availability. You can’t underwrite enough policies to make a case for it unless you are a truly huge player.”

The big Canadian players face real challenges in an increasingly concentrated domestic financial services world, Hicks says. “By centralizing their systems groups, the branch banking IT staff are likely to overwhelm the opinions of staff of other businesses by their sheer number,” he warns. “These institutions will be threatened by the high-overhead, slow-delivery and super high-availability approach that will become the default for all the various businesses. This will leave their costs high and, more importantly, their customer base open to raiding from new players offering considerably lower charges for the same services.”

As an example, Hicks cites a project at a well-known institution that will replace an existing administrative system – based on keyed-in data that is maintained on a database – with an Internet system designed for 24/7 high-availability. It is important to note that none of the information in question was required in real-time. “In a dot-com, this would have been a four to six-month effort by about three people and less than a half million dollars in hardware,” says Hicks. “However, in this organization, it has become almost a two-year effort costing many millions of dollars. They just don’t know how to think any other way.”

In an era of rapid deployment of new wholesale products over the Internet, the institution that imposes branch banking techniques on trading businesses is taking a huge risk in market share – just to gain a place on the availability curve that is largely unimportant to customers. Besides slowing time-to-market, warns Hicks, it establishes a cost structure that will not be competitive, particularly as financial services continues to deregulate and foreign players enter wholesale markets in force.

According to Hicks, institutions unwilling to accept slightly higher risks face the very real possibility that wholesale customers will move en masse to lower cost providers.


Customers expect the same service from their bank-owned broker as they get from their bank branch, yet appear reluctant to pay for it. The systems behind retail brokerage – generally built according to time-to-market wholesale standards – would require huge investments to become as robust as branch systems, driving up transaction costs substantially. Obviously, this contradicts the very reason so many clients have moved to self-serve or “discount” brokerage. On the other hand, retail customers will not move as quickly as wholesale ones for price or service considerations, so each institution must decide where they want to be on the availability curve and invest and market accordingly.

Insurance businesses provide many other points to consider, Hicks says. “The retail car insurance quote business is likely to require a different service level than the corporate health-benefit plan administrators.” People will want car insurance quotes whenever they choose, but a company benefits administrator is unlikely to work until 2 a.m. to set up a dental plan for a new employee.


So the bigger institutions face several choices. They can drive out “smart-fast” thinkers and stick to one set of standards at the peril of smothering non-branch businesses with unsupportable costs, or they can allow application groups the freedom to choose availability and risk levels – which threatens successful core branch banking values. They can also leave IT somewhat decentralized at the risk of core branch systems experts transferring to more exciting businesses that move faster to adopt new technologies and methods.

“These are hard choices that won’t be made by IT management,” Hicks says. It is the business leaders, either through active decision or decision by abdication, that will shape big investments in financial services IT over the next few years. Big investments need to be made to achieve high-availability over the Internet. According to Hicks, “These are bigger investments than old relationship businesses like, but not necessarily as big as many in the old branch businesses assume – and these decisions will determine business survival.”