Business case building for IT renewal
One by one, Australia's banks are coming to terms with the necessity of replacing heavy, complex mainframe-based architectures that date back to the 1960s with more flexible real-time processing systems.
Most banks labour under a turgid and complex architecture in which new applications have over the decades been grafted on to antiquated mainframe systems which process data overnight.
That makes it hard to compete as it takes too long to respond to customers, roll out new products and services, and process transactions.
ANZ is the latest Australian institution to declare intentions to upgrade legacy applications systems, which it will apply across its Asian operations.
It follows the Commonwealth Bank's announcement a week earlier that it would spend $580 million modernising five core systems from its legacy platform using real-time straight-through processing.
National Australia Bank is progressing down a similar path.
The ANZ's move mirrors similar initiatives in Asia where the key driver for many banks is to boost business expansion or support acquisitions in the region.
A report from industry analyst Gartner notes that international players converging on markets traditionally serviced by domestic banks have found it difficult to leverage economies of scale because of challenges such as inflexibility, long development cycles and undocumented modifications associated with legacy software.
ANZ's chief technology officer, Nick Dean, has said the bank's new platform, Infosys's Finacle core banking system, will help it to compete in priority segments and to attract new customers in Asia by speeding up delivery cycles on new services and products.
But not all banks have the same needs.
At Westpac, St George and the Australian operations of ANZ, systems are not enough of a problem to spur major modernisation projects.
"In some cases, such as Westpac, the difference in urgency comes from the fact that they remediated key bottlenecks earlier in areas such as processing platforms and that has seen business value generated," says Pascal Gautheron, head of financial services in Asia Pacific for Accenture, which is working with software vendor SAP on CBA's project.
But where bottlenecks persist, such banks plan to extend the life of key systems using new technologies such as services-oriented architecture in which re-useable components are shared.
However the cost, disruption and potential risk incurred by these projects is also a deterrent.
Westpac nurses unsettling memories of its troubled $200 million CS90 core systems project in the early 1990s, which was eventually abandoned, while the Bank of Queensland suffered numerous problems after it implemented its $40 million core banking system in 2004 in areas such as internet banking and point-of-sales systems. BOQ sourced its system from FiServ.
With modern replacement projects priced between $500 million and $800 million, any new projects will not be taken lightly.
Nevertheless, technology companies such as SAP, Oracle, EDS and specialist core banking system developer Financial Network Services have been actively pursuing the legacy replacement business, estimated to be worth around $8 billion.
They point to cost savings generated by such projects, which when done correctly, can represent as much as five to eight per cent of the total operating budget of the bank, according to research company, Celent.
Gartner says that the first wave of renewal in the Asia/Pacific among mid-tier banks has delivered high levels of return on investment, some in less than 18 months and many in less than three years.
It is still early days for Australian banks, which are still scoping out their replacement plans, but these initiatives are expected to turn around diminishing IT efficiency ratios.
Nevertheless, industry analysts are sceptical that the big banks will be able to apply real time transaction processing to high volume transaction systems such as general ledger, withdrawals and cheque transaction processing.
"The big unknown is whether the mainframes have the capacity even to handle high volume transactions online," says Alan Hansell, an advisor at industry analyst IBRS.
Big banks in Australia manage around seven million customers that generate high volumes of transactions across multiple channels that are fed into customer information systems to provide another view into the data.
It's much more manageable for smaller banks to process such activities in real time.
Hansell notes that the regional Bendigo and Adelaide Bank is among the few in the Asia Pacific to run online real time processing across the enterprise, having implemented the technology separately prior to the merger.
The bank has 1.3 million customers.
"What is unique about what it did was that it was the first to go online real time so that its data is the one version of the truth whereas with other banks, transactions sit in a shadow account until night, when it is processed," says Hansell, who has worked as a consultant to Adelaide Bank.
"Major banks only do real time processing for gross settlements transactions - everything else is batch processed overnight."
That puts big banks at a competitive disadvantage but it's not something they can easily address.
"At the moment the big banks are only ready to tackle systems at the periphery of the legacy core banking system because they don't know whether the software can handle high volume transactions," says Hansell.
Industry specialists concede there are other unanswered questions that need to be determined before replacement projects can go ahead.
As Australian banks progress down this path, a critical question will be how far they want to go - do they want to change all of the bank's architecture to fully take advantage of new online real time applications and processing or just change some of it. And which ones do they give priority to.
For example, if an ATM that traditionally functions on a nightly batch processing cycle is moved to real time data processing, banks need to weigh up which transactions are moved onto real time and which will remain batch.
As well, they will need to examine the way transactions - such as the way accounts are structured - are accomplished, says Accenture's Gautheron.
"When business asks for a new product, IT feels the pain because it has to make code changes to legacy programs and make it available across multiple channels."
Because most new products call for real time capabilities over the internet, IT has to create middleware that mediates between the applications program and the network.
"That spawns a whole lot of questions around hardware and infrastructure," says Gautheron.
Graham Rothwell, head of client management and technology operations at NAB, who was appointed earlier this year to the board of Storage Network Industry Association, says more questions spin around data migration.
Migrating the data from older mainframe-based systems to a completely new architecture is both risky and time consuming - a task that calls for system integrators to re-structure the data so that it is compatible with the new system, fill in gaps in data and rigorously test the cleansed data before turning off the old system.
"When banks go through these massive core systems projects you can't just turn the old systems off.
"They will have two systems running concurrently for some period of time and whilst this is happening they will need to ensure that they are effectively moving the data between the systems.
"This creates a lot of complexity," says Rothwell.
Another issue is whether they will be obliged to migrate all of their old data archives to the new platform.
They also need to consider peripheral systems.
"Most banks have augmented their old systems with all sorts of peripheral systems such as customer relationship management systems, data warehousing and so on. So will there be an encompassing consolidation of all these peripheral systems as well?"
Analysts concur that the banking industry is likely to forge various solutions to address these issues.
Those banks with more time on their hands will be watching the outcome with interest.