01 February 2007 Features

 

Taming the Technology

Gary Flood

Without a doubt, serious ongoing investment in technology is a given in the financial services sector. The problem is — which technology?

A hot list of continually name-checked technologies that we are told are on the minds — and indeed shopping lists — of financial sector CIOs includes service oriented architecture, virtualisation, grid computing and extensive use of blade servers in the data centre. Also on the menu would be real-time networking and messaging technologies, pivoting on the constant aim of reducing latency, as well as the usual exotic mix of algorithms being thrown at complex trading problems.

This at least would surely be representative of any conversation an IT manager will have had with a supplier in the last 18 months. “There has been a significant growth in terms of processing requirements,” says Syed Rizvi, UK managing director of Azul Systems, a firm that supplies specialist appliances for traders. “The old way of doing things is seen as no longer good enough. With the globalisation of markets, the increasing complexity of products and the need to bring them quickly to market, in either retail or trading, just adding more servers to the datacentre is seen as not a serious solution any more.”

Thus the interest in techniques that can make life more interesting. SOA is hailed as not only the best way to finally make applications talk the same lingua franca, it is regularly cited as the final “real way” to unpack all that core banking legacy functionality. Virtualisation is being touted as the optimum way to make better use of hardware resources by making actual usage and performance operate more rationally. Grid computing, we are told, is being investigated for similar reasons — to turn cheap farms of Intel processors into the equivalent of heavy metal mainframes, and if it isn’t standard PCs in that mix, step forward blade computing, where pizza box style racking and stacking is delivering cheaper but faster computing than the bad old days ever could.

Forgive the note of scepticism that may have crept into that last description, because the reality — as always with IT? — is far more complex. Some of the technology being talked about is far from tamed, and the adoption curve is far from being a straight line. Partly this is down to the nature of the market itself, of course.

“We’ve had client/server, then market trading systems, then magic middleware,” points out Kevin Covington, executive vice president of business and product development at BT Radianz. “There was also a fad for a while around workstations. The point is that there is always a popular trend that gets a lot of buzz but then something else comes along that is supposedly better.”

SOA would be a good place to start examining this proposition. The verdict on its actual day-to-day is very mixed. “SOA is a constant subject of conversation and everyone is asking about it and everyone says they want it,” says Mark Gunning, group strategic director at Temenos, the Swiss-based supplier of core banking packages. “Actual adoption is a very different picture.”

“Surveys regularly show that SOA is being talked about a lot but there is very little actual activity,” agrees David Bradshaw, principal analyst with UK analyst group Ovum. “It is seen as one of the key technologies in the innovation space, certainly, but a lot needs to happen first.”

The issue here is that a lot of the original promise of SOA cannot be delivered until extensive work is carried out on those heavy-duty back-office systems so that business logic can be more easily accessed. But as Bradshaw points out: “That means looking at stuff that’s been running on its own for quite a while and many people are extremely leery about touching it in case it breaks.”

“Wachovia in the US is still probably the most advanced deployment of any sort of SOA so far,” believes Ronan Bradley, an analyst at IT consultancy Lustratus. “What seems to be happening is that IT departments see that something like SOA could help cut down all that time spent on custom development, which still takes too long and costs too much, but they don’t want to go in for the mega-project big bang SOA implementation. So we are seeing it coming in much more on the back of projects, especially around J2EE and .Net installations, but there it’s to solve a tactical problem, usually.”

Examples of such projects include single-sign on and some decision-based services in the experience of Julian Fowler, head of market development at ILOG. “Some customers we have worked with tell us they can have up to 30 different applications needing to be combined to make such things run: obviously SOA would make that simpler. But to do that we have to really start cutting up the big old systems and that just hasn’t happened yet.”

If SOA is still in pilot mode, what about some of the other Top Ten chart technologies? Virtualisation, it turns out, is probably the most successful. “That’s really getting a lot of uptake,” says Ovum’s Bradshaw. “There is clear ROI and a lot of sense in more centralisation of operations. It’s becoming a key technology to get better utilisation of IT assets.”

That’s not a universally held conclusion. “I don’t think virtualisation has fully made it out of the labs,” says Graham Drury, senior business technologist, financial systems, at software giant CA. Indeed Richard Garsthagen, European technical marketing manager at VMWare, the leading non-Microsoft supplier of virtual machines, notes that: “Ultimately virtualisation is just a different architecture to standard physical computing. The real pay-off is going to be when applications start taking advantage of this way of working, which we are now just starting to see.”

Grid computing is similarly a mixed picture in terms of enthusiasm and adoption. “In areas like brokerage and other volatile businesses, grid is being seen as a way to dramatically reduce cost by market leading organisations,” says Temenos’ Gunning. “People like the idea of getting as much computing power as they really need.”

“This is coming along, yes, but the technology is a bit trickier to implement than something like virtualisation, and there is also investment cost to get going here,” counters Ovum’s Bradshaw. “Grid is good for raising utilisation rates but will depend on SOA to get the best out of traditional legacy servers,” thinks Azul’s Rizvi.

Blade computing overall seems to fall into the same category — great interest and some take up, but it is far from being a mainstream technology in the 2007 data centre. “They are being adopted but they bring problems as well as benefits,” claims CA’s Drury. “These include all the well known issues around power management and cooling. These are well understood issues and I’d expect to see much more use of blades on the next two years, especially for new data centres.”

So are we taming the technology? The verdict has to be that there is a lot of promise out there but it’s far from the point that we can regard things like SOA and grid as accepted, mainstream issues. The reality is that CIOs are far more likely to be spending their time “keeping the lights on”, to use the cliché. “IT managers in this sector were being told as recently as last year to cut 30% out of their operating budgets,” says Bradshaw. “That may have eased off slightly but the onus is still on cost control as opposed to raw innovation.”