Legacy Concerns Risk Overriding Architecture Decisions at Banks

"Over the past few years, with the economic downturn, the direction has been to use what you have and to squeeze the most out of it around the shop until it breaks," says Victor Ekong, senior vice president, Citi Market Data Services Asia-Pacific at Citigroup. "I think we've reached the point, now, where hoping that nothing breaks has become a risk in itself, because we're hitting the end of the lifespan for the hardware and the software."

Ekong says that although this provides opportunities for IT to present a case to the business side for replacing hardware, if long-term savings can be demonstrated and short-term costs explained, having to react to foundering technology creates issues in the way that purchasing decisions are made, particularly if replacing certain aspects becomes critical due to ongoing failures.

"The challenge there is that the end-of-life risk is now dictating that you have to do something and upgrade," he explains. "That moves away from our logic where you start from a higher level and define your stack, then work iteratively towards building a foundation which, over time, you can plug components into."

Technical Debt
Other panelists explored the concept of legacy and end-of-life systems further, with some saying that the nomenclature in itself created a set of issues for technologists.

"If you've branded a platform as legacy, then it becomes the root of all technological evil in an organization," says Stuart Gurr, head of global trading technology and IT infrastructure programs, Asia-Pacific at RBS. "We brand it technical debt, and for me, you need to manage it and get it out. If it's no longer a part of your core technology platform going forward, it becomes complicated to support, it invariably becomes more fragile and systems start going down, it consumes resources and drags the organization down."

"If you've branded a platform as legacy, then it becomes the root of all technological evil in an organization. We brand it technical debt, and for me, you need to manage it and get it out.

If branding something as legacy is a concern, though, what actually constitutes legacy itself is being re-examined in light of cost constraints. For Miguel Ortega, market data services manager at Deutsche Securities, this involves elements of considering what applications actually demand in terms of performance to run according to need. Not everything, for example, requires low-latency infrastructure, and cost savings can sometimes be achieved by thinking about changing hardware and software in a different way.

"A lot of technology works well today, and doesn't require faster machines or a lot of memory, it just works," Ortega says. "But sometimes these are big machines that take a lot of datacentre space and require a lot of cooling. Removing legacy doesn't always mean changing the way you do things; you can use the same processes, but you can move from a larger machine to a much smaller one that has similar capabilities, which will just give you a much more efficient datacentre."

The panel was moderated by James Rundle, deputy editor, sell side at Waters, and was held at the Regent Hotel in Singapore.