Reducing latency in financial systems

With the right architecture we could see the end of batch processing and the beginning of real-time information writes Husain, a consultant at Compaq Computer Asia Pacific.

Enterprises are beginning to recognize the need to integrate disparate applications for a variety of reasons, and not all of which are IT-related. Many of these enterprises are seen to be
re-engineering their business processes to accommodate customer-focused views that require close co-operation between previously stand-alone applications.

Maintaining disparate or even duplicate databases owned by each application wastes time and creates problems. To accelerate the problem, older systems do not communicate effectively with new systems; mergers and acquisitions create incompatible technology; and everyone and everything must wait until transaction and data processing are completed before anything else can be accomplished.

For example, the problem arises when a financial institution runs on several databases, and several transactions in a single account are not processed until the end of the day. As a result, the same money may have already been approved for different users, over-extending available resources and increasing the time spent correcting mistakes.

With the same scenario applied to the retail sector, a lack of integration can also translate to undetected losses as small as fraudulent returns or as large as outright fraud. From health care to telecommunications to transportation, there is hardly an industry that would not profit from increasing the integration of information or eliminating the latency in its processes.

At the moment, many enterprises have adopted an Enterprise Application Integration (EAI) strategy that can effectively cluster dozens or hundreds of complex applications. However, difficulties still exist. The very nature of integrating so many applications with EAI can make those applications more susceptible to failure. When one linked system goes down, it creates a domino effect as systems are not synchronized and not operating in real-time but in periodic batch updates.

Gartner's Roy Schulte, vice president and research fellow, developed the term "zero-latency strategy" as "any strategy that exploits the immediate exchange of information across technical and organizational boundaries to achieve business benefit."

An enterprise that executes such a strategy can then be labelled a Zero Latency Enterprise (ZLE) - one that integrates all of its systems into an end-to-end enterprise architecture. This is independent of platform and coding - whether they are legacy, client-server or even web-based. A ZLE architecture is also capable of linking customers, suppliers, partners and service providers together, while maintaining security and control.

In short it enables companies to gather and process information continuously, in real-time and without delay. This is to ensure that information is available immediately across the entire enterprise and for leverage to serve customers and solve critical challenges.

ZLE is also a key factor for application integration. By channeling information through the enterprise, from one application to another, and with near zero latency, it is possible to streamline business processes, reduce costs and respond more quickly to marketplace demand and competitive pressures.

In short, enterprises that are able to respond within a short timeframe have a competitive advantage; they will be in a position to sell more products and services, deliver better customer service with greater efficiency and most importantly be able to leverage on business opportunities for more effective marketing initiatives.

A ZLE initiative needs an operational data store (ODS) server that integrates multiple applications and data sources in real-time implementation. The ODS is an enterprise-wide, real-time data cache that ensures if one application goes offline, the remaining applications maintain access to its data. Caching the data in the central repository loosens application dependencies without breaking integration bonds. IT provides for business continuity, as faults in one application do not create a domino effect. With integrated ZLE, an update of any one application also updates applications downstream.

The data integration layer, enabled by the ODS server, is capable of supporting business intelligence functions, as well as EAI real-time data inserts needed for many 24 x 7 operations, that are essential for the financial industry. It can support simultaneous online transaction processing and business intelligence workloads through prioritized mixed workload functionality. In addition, the data integration layer is also well integrated with ZLE independent software vendor partner applications that provide data profiling; data cleansing; campaign management; data mining; and rule-based, real-time informed decision-making capabilities.

When a company becomes a zero latency enterprise, it can act immediately on up-to-the-second business and customer information for competitive advantage. Real-time information gives companies the power to see their business as it is right now, know their customers as they are right now, and act on that up-to-the-second information without delay: they can see, know and act now.

It is estimated that the deployment of a zero latency strategy could reduce information-processing costs from anywhere between five per cent to 35 per cent. In addition, zero latency reduces the number of applications in a customer's profile by allowing a single clip-on application to perform the work currently done by many applications.

It is believed that a zero latency approach could achieve the following:

- Reduce churn from 15 to five percent
- Lead to an improvement in profitability by six percent to 15 percent
- Improve customer satisfaction ratings by 10 percent to 20 percent

The main challenge in implementing this type of architecture is usually not a technical one. It is the co-operation of the many parts of the organisation to see how this would benefit the entire enterprise.

Let us take a typical bank as an example, there would be a core banking department, a retail banking department, a loans department and so on. For true CRM (Customer Relationship Management) to be offered, access to all data channels will be required. The solutions that most banks purchase today are bought to achieve a single aim, and are not designed to deliver a true enterprise solution.

Thus, most departments are reluctant to share data and systems, especially in real-time. Most will allow access to their data (say an ATM transaction) as an overnight batch run. However, this does not allow an enterprise to make a real-time offer, because the customer interactions do not get to the business intelligence ZLE engine in real-time. The technology already exists, however, in common standards such as CORBA or Tuxedo, and common formats such as XML.

Typically most organizations do not look at what a true enterprise solution (involving many platforms, applications and standards) really is. Most departments are cost-driven or profit centres in their own domain (credit card systems, loans, etc.), so they do not think of what effect their system has on another department. The challenge is often the lack of vision of the whole enterprise, and why an open, scalable and reliable architecture is essential to achieve true market differentiation.

According to Ross Altman, Research Director at Gartner, information that is fresh has the most value. In real-time ZLE, information is delivered to the right place, at the right time for maximum business value. Organizations are moving towards reduced latency environments; it is just a question of when. The longer one waits, the worse it gets. The sooner an enterprise embraces and implements the concept of zero latency, the stronger its competitive position is likely to be and the happier customers will be.

Article limit is reached.

Hello! You have used up all of your free articles on FinanceAsia.

To obtain unlimited access to our award-winning exclusive news and analysis, we offer subscription packages, including single user, team subscription (2-5 users), or office-wide licences. To help you and your colleagues access our proprietary content, please contact us at [email protected], or +(852) 2122 5222